Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials 9780841227309, 9780841227316

Table of contents :
bk-2012-1103_cover......Page 1
bk-2012-1103.fw001......Page 2
bk-2012-1103.pr001......Page 7
bk-2012-1103.ot001......Page 10
bk-2012-1103.ch001......Page 11
bk-2012-1103.ch002......Page 32
bk-2012-1103.ch003......Page 59
bk-2012-1103.ch004......Page 82
bk-2012-1103.ch005......Page 97
bk-2012-1103.ch006......Page 128
bk-2012-1103.ch007......Page 135
bk-2012-1103.ch008......Page 147
bk-2012-1103.ch009......Page 159
bk-2012-1103.ch010......Page 170
bk-2012-1103.ch011......Page 187
bk-2012-1103.ch012......Page 199
bk-2012-1103.ch013......Page 220
bk-2012-1103.ch014......Page 231
bk-2012-1103.ch015......Page 240
bk-2012-1103.ch016......Page 249
bk-2012-1103.ch017......Page 282
bk-2012-1103.ix002......Page 297

Citation preview

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.fw001

Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.fw001

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

ACS SYMPOSIUM SERIES 1103

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.fw001

Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials Patricia L. Lang, Editor Ball State University Muncie, Indiana

Ruth Ann Armitage, Editor Eastern Michigan University, Ypsilanti, Michigan

Sponsored by the ACS Division of Analytical Chemistry

American Chemical Society, Washington, DC Distributed in print by Oxford University Press, Inc.

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.fw001

Library of Congress Cataloging-in-Publication Data Collaborative endeavors in the chemical analysis of art and cultural heritage materials / Patricia L. Lang, editor, Ball State University, Muncie, Indiana, Ruth Ann Armitage, editor, Eastern Michigan University, Ypsilanti, Michigan ; sponsored by the ACS Division of Analytical Chemistry. pages ; cm. -- (ACS symposium series ; 1103) Includes bibliographical references and index. ISBN 978-0-8412-2730-9 (alk. paper) 1. Spectrum analysis--Congresses. 2. Cultural property--Conservation and restoration-Congresses. I. Lang, Patricia L., editor of compilation. II. Armitage, Ruth Ann, editor of compilation. III. American Chemical Society. Division of Analytical Chemistry, sponsoring body. QD95.C617 2012 701′.5435--dc23 2012018630

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, ANSI Z39.48n1984. Copyright © 2012 American Chemical Society Distributed in print by Oxford University Press, Inc. All Rights Reserved. Reprographic copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Act is allowed for internal use only, provided that a per-chapter fee of $40.25 plus $0.75 per page is paid to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. Republication or reproduction for sale of pages in this book is permitted only under license from ACS. Direct these and other permission requests to ACS Copyright Office, Publications Division, 1155 16th Street, N.W., Washington, DC 20036. The citation of trade names and/or names of manufacturers in this publication is not to be construed as an endorsement or as approval by ACS of the commercial products or services referenced herein; nor should the mere reference herein to any drawing, specification, chemical process, or other data be regarded as a license or as a conveyance of any right or permission to the holder, reader, or any other person or corporation, to manufacture, reproduce, use, or sell any patented invention or copyrighted work that may in any way be related thereto. Registered names, trademarks, etc., used in this publication, even without specific indication thereof, are not to be considered unprotected by law. PRINTED IN THE UNITED STATES OF AMERICA In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.fw001

Foreword The ACS Symposium Series was first published in 1974 to provide a mechanism for publishing symposia quickly in book form. The purpose of the series is to publish timely, comprehensive books developed from the ACS sponsored symposia based on current scientific research. Occasionally, books are developed from symposia sponsored by other organizations when the topic is of keen interest to the chemistry audience. Before agreeing to publish a book, the proposed table of contents is reviewed for appropriate and comprehensive coverage and for interest to the audience. Some papers may be excluded to better focus the book; others may be added to provide comprehensiveness. When appropriate, overview or introductory chapters are added. Drafts of chapters are peer-reviewed prior to final acceptance or rejection, and manuscripts are prepared in camera-ready format. As a rule, only original research papers and original review papers are included in the volumes. Verbatim reproductions of previous published papers are not accepted.

ACS Books Department

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.pr001

Preface The chemical analysis of art and cultural heritage materials began two centuries ago. In 1815 renowned British chemist Sir Humphry Davy described the analysis of pigments on objects excavated from the ruins of Pompeii in a paper that he read to the Royal Society (1). He wrote: “When the preservation of a work of art was concerned, I made my researches upon mere atoms of the colour, taken from a place where the loss was imperceptible: and without having injured any of the precious remains of antiquity, I flatter myself I shall be able to give some information, not without interest to scientific men, as well as to artists, and not wholly devoid of practical applications.” Sir Davy hoped to not only become acquainted with the nature and chemical composition of the pigments, but to discover some idea of the manners and styles of the artists (2). The scientists authoring the chapters in Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials have taken the same footpath as Sir Davy in regard to the practicality of their research, but they have outpaced Davy in its appeal to a broader audience. The reader will find interesting chapters describing: the process of uncovering forgeries and counterfeits (Chapters 1, 11, 12, 16); the pedagogy of teaching the chemical analysis of art to undergraduates and the history of that “movement” (Chapters 13, 14, 15); the results of scientific investigations on art and cultural objects that have been performed primarily by students and their faculty mentor (Chapters 10, 11, 16, 17); the use of the latest technology in identifying pigments on prehistoric rock paintings, the dating of ancient objects, or the characterization of dyes or biomarkers on archeological samples (Chapters 4, 5, 6, 7, 8). The reader will also enjoy reading the viewpoint of museum conservators who have played a major role in writing and contributing to the science reported in some of the chapters (Chapters 1, 2, 3, 12 and 16). Perhaps most thought-provoking, is a chapter in Collaborative Endeavors that asks the question, “What can science alone tell us?” (See Chapter 9.) But the book is not just a collection of several case studies of describing the chemical composition of objects of cultural or artistic interest; the book aims to illustrate how the chemical and physical analysis of art and cultural heritage materials is a perfect model of collaboration with museum curators, with historians, with students, with religious scholars, anthropologists, and/or with other specialists who partner to answer interesting and important questions about an archeological work or piece of art worthy of study: What are the materials? How was it made? Who influenced the work? How has it changed or deteriorated? Why was it made? Since no one scholar or scientist can answer all these questions, experts from many areas using many different kinds of analytical techniques ix In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.pr001

are drawn together in Colloborative Endeavors to share their knowledge and experience. As a result, an understanding of how the molecular and atomic world plays a role with physical products of human expression is presented from many different perspectives. Sir Davy was not so lucky when it came to cooperative efforts. In 1821 he read before the Society, “Some Observations and Experiments on the Papyri Found in the Ruins of the Herculaneum” (3). He wrote: “I should gladly have gone on with the undertaking, from the mere prospect of a possibility of discovering some better results, had not the labour, in itself difficult and unpleasant, been made more so, by the conduct of the persons at the head of the department in the Museum….and these obstacles were so multiplied, and made so vexatious towards the end of February, that we conceived it would be both a waste of the public money, and a compromise of our own characters, to proceed.” Sir Davy’s experience emphasizes the absolute necessity for cooperative efforts between various scientific, community, and/or academic units in solving these intriguing mysteries. However, what is different between Davy’s era and the present (except the obvious advances in technology) is the network of information and education in regard to the analytical process of art and cultural material analysis. Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials shows what can be accomplished as a result of that network. There are at least 32 museums, universities, and agencies, who helped directly to make the projects presented within possible. They include but are not limited to: Chemistry Department, Eastern Michigan University Archaeogrpahics, Moscow, ID Archaeological/Historical Consultants, Oakland, MI Conservation Department, Detroit Institute of Arts Department of Chemistry, Washington and Lee University University Collections of Art and History, Washington and Lee University Balboa Art Conservation Center The San Diego Museum of Art Timken Museum of Art Collections, Mount Vernon Estate and Gardens Smithsonian National Portrait Gallery Smithsonian American Art Museum J. Paul Getty Museum Royal Picture Gallery Maurishuis, The Netherlands Scientific Research and Analysis Laboratory, Winterthur Museum Department of Chemistry, Rhodes College Department of Chemistry and Biochemistry, University of Mississippi Department of Mathematics and Computer Science, Rhodes College McCrone Associates, Westmont, IL Book and Paper Conservation, The Walters Art Museum, Baltimore, MD New Testament & Early Christian Literature Department, University of Chicago Regenstein Library, University of Chicago Department of Chemistry and Biochemistry, University of Detroit Mercy Department of Chemistry, Millersville University x In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.pr001

Department of Chemistry Whitman College Department of Chemistry, Ball State University David Owsley Museum of Art, Ball State University Chemistry Department, University of West Georgia Chemistry Department, Clark Atlanta University Anthropology Department, Hofstra University Indianapolis Museum of Art Conservation Department, Buffalo State College The authors of this book have been a delight with whom to work. One of the insights gained in editing a book is to see the response of the authors after the reviews come back. All have worked hard and have been gracious in providing revisions to their work, when necessary, in order to made the science presented as clear and accurate as possible. They are the experts in their field. Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials is the result of the Chemistry of Art Symposium held at the 2011 Central Regional Meeting of the American Chemical Society at IUPUI in Indianapolis, IN on June 9th and organized by P. Lang. The editors would like to gratefully acknowledge Dr. Corinne Deibel, Professor of Chemistry, Earlham College, for her organization of that meeting and support of the symposium. We also thank the ACS editors, especially Nikki Lazenby for her thoughtful assistance with this book, and a big thanks goes to the many colleagues who reviewed the book and its chapters.

References 1.

2. 3.

Some Experiments and Observations of the Colours used in Paintings by the Ancients. In The Collected Works of Sir Humphry Davy; Davy, J., Ed.; Smith, Elder and Co. Cornhill: London, 1840; Vol. VI, pp 134−135. Probing the Authenticity of Antiquities with High-Tech Attacks on a Microscale. Science 1988, 239, 1374. The Ruins of Herculaneum. In The Collected Works of Sir Humphry Davy; Davy, J., Ed.; Smith, Elder and Co. Cornhill: London, 1840; Vol. VI, pp 173−174.

Patricia L. Lang, Professor and Chair Department of Chemistry Ball State University, Muncie, IN 47306 765-285-5516 (telephone), [email protected] (e-mail)

Ruth Ann Armitage, Professor Department of Chemistry Eastern Michigan University, Ypsilanti, MI 48197 734-487-0290 (telephone), [email protected] (e-mail) xi In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Editors’ Biographies

Downloaded by UNIV OF GUELPH LIBRARY on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ot001

Patricia L. Lang Patricia L. Lang, Professor and Chair of Chemistry Ball State University, earned a B.S. in Chemistry from Ball State in 1983. She completed a Ph.D. in Physical Chemistry from Miami University with Dr. Jack E. Katon in 1987 on the applications of infrared and Raman microspectroscopy. Her research focuses on the spectroscopic characterization of a wide range of different materials that include asbestos, monolayers, fibers, and bacteria. She has written and presented extensively on the analysis of historic materials including parchment, paper sizing, pigments, binders, and paint additives. In her 25 years at Ball State Dr. Lang has mentored the research of over 60 students.

Ruth Ann Armitage Ruth Ann Armitage, Professor of Chemistry at Eastern Michigan University, earned a B.A. in Chemistry from Thiel College in 1993. She completed a Ph.D. in Analytical Chemistry at Texas A&M University with Dr. Marvin Rowe in 1998 on radiocarbon dating of charcoal-pigmented rock paintings. Since joining EMU in 2001, her research focus has remained on characterizing archaeological and cultural heritage materials including rock paintings, residues, and colorants in textiles and manuscripts in collaboration with archaeologists and museum conservation scientists, as well as both undergraduate and graduate students.

© 2012 American Chemical Society In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Chapter 1

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

What’s Wrong with this Picture? The Technical Analysis of a Known Forgery Gregory D. Smith,*,1 James F. Hamm,2 Dan A. Kushel,2 and Corina E. Rogge2 1Indianapolis

Museum of Art, 4000 Michigan Road, Indianapolis, IN 46208 2Art Conservation Department, 1300 Elmwood Avenue, Buffalo State College, Buffalo NY 14222 *E-mail: [email protected]

Robert Lawrence Trotter was convicted in federal court in 1990 of producing and selling fake American primitive style folk art. A methodical ‘reverse engineering’ of one of his confiscated paintings, Village Scene with Horse and Honn & Company Factory, was undertaken to determine what telltale signs might exist to identify this work as a forgery. Currently 39 other Trotter fakes are yet unaccounted for and are potentially circulating on the art market or belong to private or institutional collections. A crescendo approach to the critical examination of this painting began with a simple visual assessment followed by diagnostic imaging using X-ray, near infrared, and UV radiation sources. Non-sampling chemical analysis with X-ray fluorescence and Raman microspectroscopies was conducted next to determine specifically the forger’s materials. Finally, additional information was gathered from invasive sampling approaches including cross section analysis, FTIR microspectroscopy, and pyrolysis-gas chromatography-mass spectrometry. Copious clues to the work’s inauthenticity exist at every level of investigation. Although simple visual examination would raise questions as to the artwork’s genuineness, diagnostic imaging and chemical analysis prove beyond a doubt that the work is a modern fake. Anachronistic pigments and improbable construction techniques are evidence that this is not an authentic piece of 1860s folk art. © 2012 American Chemical Society In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Introduction In 1990, Robert Lawrence Trotter was sentenced to ten months in federal prison for a decade long scheme that involved the production and sale of fake American 19th-century primitive style paintings (1–3). By his own admission, Trotter conducted fifty-two sales of his fakes and forgeries from 1981 to 1988 involving six art dealers and twenty-nine auction houses in eleven states. His ill-gotten gains were in excess of $100,000 (1), although some of his earliest fakes sold for paltry sums, one for only $36 (1, 2). The actual crime to which he pled guilty was wire fraud related to these transactions (2). Trotter’s familiarity with the art and antiques market in the early 1980s made him aware of the innumerable fakes that exist, especially in the primitive or folk art style. Being an amateur artist himself, he joined in the production of bogus artworks in order to augment his income. Initially these generic, anonymous folk art pieces fetched only modest sums at auction and attracted little attention. His works included typical folk art scenes and sitters, and he utilized a pastiche of the physical characteristics admired in many styles of folk art to enhance the appeal of his forgeries. In 1988 Trotter, being impatient with the trivial earnings of his fakes up to this point, left the relatively safe confines of low value, anonymous primitive art and began directly imitating the styles of well-known 19th-century folk artists such as M. W. Hopkins, Ammi Phillips, and Noah North, and finally the artist best known for his trompe l’oeil paintings, John Haberle. These more ambitious attempts came to the attention of the community of scholars and collectors specializing in these artists’ works, and their authenticity began to be questioned. Because of suspicions regarding Trotter’s fake ‘Haberle,’ the FBI established a sting operation that ultimately caught him red-handed. Dan Hingston, an auction manager caught up in the scheme, noted, “We’re lucky Trotter raised the stakes. If he’d stayed at this level [anonymous, generic folk art], he could still be doing it (2).” Of the fifty-five fakes he produced, only sixteen of these works were located by the FBI in their investigation (1). Moreover, only five of these identified works were seized by the Bureau or turned over to it as part of Trotter’s compensation settlement. One of these paintings, known as Village Scene with Horse and Honn & Company Factory, which was signed “Sarah Honn,” and dated “May 5, 1866 A.D.,” was actually painted by Trotter in 1985 and ultimately was given to the Art Conservation Department at Buffalo State College by the courts in 1991 to be used for study and research (1). This landscape painting is shown in Figure 1(a). With thirty-nine of Trotter’s fake paintings still unaccounted for and likely circulating on the art market or part of private or institutional collections, the Buffalo State College faculty decided to undertake a comprehensive study of the work to determine what signs might point a conservator or curator to question its authenticity. This is particularly important since the current condition of the Village Scene painting is poor and has noticeably worsened in the intervening years, likely the result of the techniques used by Trotter to enhance its aged appearance. Based on this observation, it is reasonable to assume that others of Trotter’s oeuvre will likely be brought to conservators for stabilization, cleaning, 2 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

and treatment if they haven’t already been restored – many of his works were immediately taken by their new owners to restorers (2). This investigation brought together the combined expertise of paintings conservators, conservation scientists, and imaging specialists. The approach to the examination began with the simplest form of exploration, namely close critical observation of the painting’s composition and obvious physical construction. This level of investigation is available to all conservators, curators, and collectors. Next, macroscopic imaging techniques familiar to art conservators and forensic investigators were used to capture images of the artwork using X-ray, near-infrared, and UV radiation sources. These images provide complementary information on the artist’s working methods, the condition of the artwork, and hidden aspects of its construction. Finally, scientific analysis, first using only non-destructive methods and later techniques that required sampling from the artwork, were utilized to explore the materials used by the forger and to compare them to what would be expected for a true 1860s folk art painting. This level of investigation is only possible at the most technologically sophisticated cultural heritage institutions, although motivated clients could arrange for contract analysis of their paintings.

Experimental Section Imaging Techniques Color photographs were acquired using a Sinar view camera with a Better Light digital scanning back CCD trilinear array with 3200 K incandescent illumination. Ultraviolet-induced visible fluorescence images of the painting were captured with a Nikon D100 digital camera (CCD array) with Wratten 2E and CC40Y filters. The camera was adjusted to white balance 7000 K and Adobe Camera Raw® tint +11. The source of UVA radiation (315 to 400 nm) was a pair of high-pressure mercury lamps filtered of their visible emission lines. A near infrared (NIR) image in transmission mode was acquired with the digital scanning back mentioned above, but modified with a Wratten 87C visible blocking filter, thus restricting the camera sensitivity to 850 to 1000 nm. The painting was exposed to NIR radiation from incandescent photo lamps directed onto the verso with the camera capturing the radiation transmitted through the painting. A radiograph of the painting was recorded on Kodak Industrex Rapid 700 radiographic paper. The Philips X-ray tube voltage was 30 kV and exposure was 525 mA•sec at a 60 in. film-focus distance. The radiograph of the experimental canvas mock-up discussed below was recorded on Kodak Industrex M100 film using the same experimental parameters. Digital versions of all images were adjusted in Adobe Photoshop for color correction, tint, exposure, mosaicking, and sharpness as necessary. Microfocus X-ray Fluorescence (XRF) X-ray fluorescence spectra were collected using a Bruker ARTAX energy dispersive X-ray spectrometer system. The excitation source was a molybdenum 3 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

target X-ray tube with a 0.2 mm thick beryllium window, operated at 50 kV and 600 mA current. The X-ray beam was directed at the painting through a masked aperture of 0.65 mm diameter. X-ray signals were detected using a Peltier cooled XFlash 2001 silicon drift detector. Helium purging was used to enhance sensitivity to light elements. Spectra were collected over 60 sec live time.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Raman Microspectroscopy Raman spectra of pigments were acquired using a Bruker Senterra microscope suspended on a Z-axis gantry. The ‘Z-stage’ allowed the entire artwork to be placed directly under the 50X ultra-long working distance objective of the microscope. Excitation at 785 nm and 1.1 mW power at the laser focus was used to stimulate Raman scattering from an area of approximately 1-2 μm diameter. To reduce the interference due to fluorescence, an area of agglomerated pigment particles was chosen in an exposed fissure in the paint film. The resulting spectrum was measured at 3-5 cm-1 spectral resolution with several hundred seconds of spectral coaddition. The pigment’s identity was ascertained by comparing its spectrum to those of likely reference materials. Sampling and Cross Section Preparation Paint samples were acquired from the painting under a stereomicroscope at low magnification using chemically etched tungsten needles and a surgical scalpel. Sampling was limited to existing areas of damage, abrasion, or cracks. Disperse samples of pigments and media were collected as surface scrapings or small paint flakes from a specific area or paint passage, and these were stored on glass well slides under cover slips until analyzed. Cross section samples were acquired by cutting vertically through the varnish, paint, and ground layers to the underlying canvas substrate using a 500 μm tip microchisel (Ted Pella). The sectioned sample was generally less than 100 μm in the long dimension. These samples were mounted in Ward’s Bio-plastic™ polyester resin. Once the embedding medium fully cured, the plastic block was cut and polished on a series of Micromesh™ cloths to expose the painting’s cross section. Darkfield images of the sectioned samples were acquired on a Zeiss AxioImager A1m compound microscope with a 20X objective using an MRc5 digital photomicrography camera. The same area was then examined under UV irradiation for signs of visible luminescence. A DAPI filter cube set allowed narrowband excitation between 325 and 375 nm with observation throughout the visible spectrum (λ > 412 nm). Fourier Transform Infrared (FTIR) Microspectroscopy Infrared spectra were collected using a Continuum microscope coupled to a Magna 560 FTIR spectrometer (Thermo Nicolet). Samples were prepared by flattening them in a diamond compression cell (Thermo Spectra Tech), removing the top diamond window, and analyzing the thin film of sample in transmission mode on the bottom diamond window. An approximately 100 μm square 4 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

microscope aperture was used to isolate the sample area for analysis under a 15X Schwarzschild objective. The spectra are the average of 32 scans at 4 cm-1 spectral resolution. Correction and subtraction routines were applied using the instrument’s Omnic software as needed to eliminate interference fringes, sloping baselines, or peaks from interfering spectral components. Sample identification was aided by searching a spectral library of common conservation and artists’ materials (Infrared and Raman Users Group, http://www.irug.org).

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Pyrolysis-Gas Chromatography-Mass Spectrometry (Py-GC-MS) A small scraping of paint was analyzed by Py-GC-MS after derivatization of the sample using tetramethylammonium hydroxide (TMAH). The sample was analyzed using a Frontier Lab Py-2020D double-shot pyrolyzer system with a 320°C interface to an Agilent Technologies 7820A gas chromatograph and 5975 mass spectrometer detector. An Agilent HP-5ms capillary column (30 m x 0.25 mm x 0.25 µm) was used for the separation with 1 mL/min of He as the carrier gas. The split injector was set to 320°C with a split ratio of 50:1. The GC oven temperature program was 40°C for 2 min, ramped to 320°C at 20°C/min, followed by a 9 min isothermal period. The MS transfer line was at 320°C, the source at 230°C, and the MS quadrupole at 150°C. The mass spectrometer was scanned from 33-600 amu at a rate of 2.59 scans/sec with no solvent delay. The electron multiplier was set to the auto-tune value. Samples were placed into a 50 µL stainless steel Eco-cup, and 3 µL of a 25% methanolic solution of TMAH were introduced for derivatization. After 3 min the cup was placed into the pyrolysis chamber where it was purged with He for 3 min. Samples were pyrolyzed using a single-shot method at 550°C for 6 sec. Sample identification was aided by searching the NIST MS library and by comparison to pyrograms of authentic samples.

Results and Discussion Visual Examination On the surface, the painting has all the hallmarks of a highly desirable piece of American folk art from the 19th-century. The scene, Figure 1(a), is a typical primitive style landscape showing a small village surrounded by pastures. The artist appears to be an amateur based on the naïve sense of perspective. The careful observer is rewarded by recognizing a nearly obscured signature in the lower right hand corner of the painting. The autograph, in clear block lettering in brown paint, reads “Sarah Honn May 5, 1866 A.D.” The signature is interesting on several levels. First, folk art paintings are rarely signed, and this picture would be especially valuable because the artist is presumably a woman. History has recorded very few female folk art painters outside of the well known Susan Waters and Grandma Moses. When viewed under the stereomicroscope, it is obvious that the aging cracks run through the artist’s paint as well as the signature, indicating that the two were contemporary and that the entire painting with its autograph aged together. 5 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

The village scene composition might well be a nod to the historical practice of itinerant painters of the mid-19th century who traveled the country drawing or painting a client’s land holdings in return for a small payment or even a place to stay. In this instance, the presence of a small sign over the door of the red brick factory, which reads “Honn & Co.,” adds an interesting twist in that the artist is presumably recording her own family’s property. All attempts to identify a Sarah Honn living in late 19th-century America or a Honn & Co. business were fruitless. This anonymity reveals the forger’s cleverness. Sarah Honn as a person is more believable because of the existence of her place name, and yet being purely fictitious, there are no potential inconsistencies to be discovered by a studious researcher. The verso of the painting, Figure 1(b), provides additional ‘badges of authenticity’ for a gullible collector. The back of the canvas is lined onto a piece of blue-striped mattress ticking, partly obscured by a dark stain, presumably due to mold. Mattress ticking has occasionally been used as a cheap canvas material by itinerant painters and folk artists. However, in this instance the mattress ticking is not the artist’s canvas, but rather a glued relining fabric, which is unique in the experience of the authors. Relining of a worn canvas is a preservation intervention performed when the original is structurally too compromised or weakened to support the paint. Careful examination of the paint surface in raking light showed a friable, undulating paint layer, but no indication of tears or holes in the canvas that would have necessitated relining. The presence of the mattress ticking as a lining fabric is perhaps the most telling outward clue that the artwork has been at least embellished to enhance its desirability. Figure 2 shows a detail shot of the bottom tacking margin, i.e. the lower flap of canvas that is used to tack the painting onto a rectangular wooden stretcher. The upper layer of fabric, the original canvas, is frayed and reveals the blue-striped ticking underneath. The metal tacks, which were removed from their original locations (dashed circles) during the relining, present an improbable situation. The artist’s white priming layer runs over top of both the original tack holes and the tack heads in their current position, suggesting the canvas was removed from the stretcher, lined, re-tacked to the stretcher, and then primed and painted. This necessitates that the painting was relined before the surface image was painted, a situation that makes no logical sense in terms of a painter’s normal practice. Imaging Techniques UV-induced visible fluorescence imaging is a useful survey technique to gauge the condition of an artwork (4, 5). Artists’ paints and varnishes tend to develop fluorophores as they age, giving old paintings a characteristic luminescence when irradiated with long wavelength UVA lamps. Recent areas of retouching over damages or paint losses will not have had the time to develop the same level of fluorescence, thus appearing as darker patches in the fluorescence image. In describing his last criminal endeavor, Trotter mentioned that on the ‘Haberle’ painting he, “. . . used a thin coat of copal varnish. It’s browner and thinner and with a blacklight it tends to throw that even overall glow that can fool people not used to using a blacklight (2).” 6 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 1. Color images of the front (a) and verso (b) of “Village Scene with Horse and Honn & Company Factory,” 40.8 cm x 51.1 cm. Courtesy of Buffalo State College. 7 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 2. Detail of the lower tacking margin showing former tack holes (dashed circles). Courtesy of Buffalo State College.

Figure 3(a) shows a color image of the front of the Village Scene painting when exposed to UVA. It is obvious that Trotter implemented a similar coating technique, since the entire surface of the painting emits an uneven, striated cool fluorescence suggestive of a natural resin coating, although not necessarily a copal varnish, partially obscured by UV blocking dirt and grime. Conservators, however, are very used to utilizing a blacklight, and this image is startling to a highly trained eye for a painting that purports to be nearly 100 years old, has already undergone a relining, and bears extensive mold growth suggesting years of exposure to moist conditions. There are no dark patches in the fluorescence image that would indicate a history of retouching, structural repair, or restoration. Such a homogenous surface fluorescence is unusual unless a forger or dealer is trying to be duplicitous by adding an intentionally concealing surface coating. Figure 3(b) shows the same imaging technique applied to the verso of the artwork. The heavy, seemingly brush applied mold stains are faintly fluorescent, which is not atypical for molds. However, the relining fabric, best seen in the upper right corner, is far more luminescent. Optical brighteners are applied to modern fabrics or are included in laundry detergents to give textiles a ‘whiter-than-white’ appearance. Fluorescent brighteners are a post-WWI invention (6), providing a clear indication that this lining fabric was applied in the 20th-century.

Figure 3. UV-induced visible fluorescence images of the (a) front and (b) verso of the painting. Courtesy of Buffalo State College. 8 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Transmitted NIR photography relies on the transparency of many artists’ materials to long wavelength radiation thus allowing imaging of underlying structures or compositions that include infrared opaque materials (7). Most often NIR examination is conducted to visualize underdrawings or preparatory cartoons executed in graphite or charcoal. In the transmitted NIR detail photograph of the white building on the horizon, Figure 4, there is no indication of a carbon-based underdrawing. The opaque passages are simply those surface features that utilize a carbon containing black paint. It is obvious that the artist painted in the landscape prior to placing the buildings since the horizon line is clearly running behind the central white building. These construction details evidenced by NIR imaging are not an indication of fakery for a primitive-style painting since it is easy to imagine the amateur folk artist painting what they saw in a very spontaneous way without significant planning or sketching.

Figure 4. Transmitted NIR image detail showing the horizon line running through the building and the fine craquelure pattern. Courtesy of Buffalo State College. More importantly, the transmitted NIR image highlights in stark contrast the islands of paint separated by a scaly craquelure. This significant cracking and its evenness across the surface of the painting deviates from the typical age-induced crack patterns observed in old oil paintings. Naturally occurring cracks create a pattern perpendicular to radiating lines of stress originating in the restrained canvas corners, which are often exacerbated by low relative humidity or low temperatures (8). The cracking in Village Scene is similar to cracking that occurs when paint is dried by heat, causing rapid, simultaneous contraction of the entire paint surface (9). Trotter again has provided some clues to the techniques used to simulate aging. He reportedly used “. . . lots of driers [siccatives] …” and would age his finished paintings for a week under a sunlamp (2). No doubt the 9 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

chemical accelerators coupled with the heat from the lamp contribute to the small, even cracking observed in this forgery. In another work by Trotter, a conservator observed under the microscope that the age cracks had actually been scratched into the paint using a sharp stylus (2). X-radiography is often utilized by conservators to image the distribution of heavy metal pigments in a painting. Until the commercialization of a synthetic route to ultrapure TiO2 pigment in the late 1910s, the most common white artists’ pigment was basic lead carbonate [2PbCO3•Pb(OH)2] (10), being the principle white paint, but also mixed into other colors to adjust their value. As a result of this widespread use of lead white pigment, most X-ray images of historic artworks show a ghost-like image of the surface painting, but also reveal underlying artist’s changes as well as abandoned compositions or reused canvases.

Figure 5. Radiograph of “Village Scene”. Figure 5 shows the radiograph of Village Scene. It is immediately obvious that no heavy metal pigments, at least none containing lead, were used to create the surface image as there is hardly any indication of the landscape. A palette devoid of lead white is inconsistent with a 1860s provenance. Moreover, one sees only amorphous, high contrast passages with blurred, indistinct borders unrelated to any figure or structure in the surface painting. In the investigation of the Trotter case, the FBI found that the forger often visited antique shops where he would buy inexpensive period paintings (1, 2). Trotter confessed that these paintings provided him with the old canvas support necessary to produce a convincing 10 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

fake. The application of commercial paint strippers removed the painted image on the authentic canvas, allowing Trotter to prepare his fakes without any telltale underlying textures. The radio-opaque indistinct passages in the X-ray image of Village Scene suggested that remnants of an old lead white containing ground were not removed by the paint stripper, perhaps because the ground was pushed into the canvas weave. The authors produced a mock-up using period canvas with a lead white oil ground which was removed using Zip-Strip® purchased at a local hardware store. After softening the oil paint and scraping it from the canvas with a putty knife, a radiograph of the mock-up, reproduced in (11), was captured using identical instrumental parameters to those used to collect the radiograph in Figure 5. The two images show a clear similarity, thereby confirming Trotter’s reuse of canvas from an old painting for the creation of Village Scene. Noninvasive Scientific Analysis Sophisticated scientific analysis is available to many conservators and curators at larger institutions employing conservation scientists. Even at smaller institutions and in private practices, the availability of a limited number of scientific instruments, optical microscopes, and microchemical testing equipment is common. If such instrumentation is accessible, then the investigator can gain a specific knowledge of the materials used by an artist. For the sake of authentication, or more accurately ‘inauthentication,’ one is typically looking for anachronistic or otherwise undocumented materials or methods for a particular artist, style, or time period being employed in the creation of the suspect artwork. X-ray fluorescence (XRF) spectroscopy is one technique that is widely available, either as a laboratory based microanalytical instrument or as the growingly popular handheld XRF units. The power of XRF lies in its non-invasive nature and the fact that the resulting elemental spectrum can be used to infer the inorganic or organometallic pigments used by the artist. Village Scene was subjected to an exhaustive analysis of the forger’s palette using a microfocus lab-based instrument. When creating his last forgery, i.e. the trompe l’oeil style ‘Haberle’ painting, Trotter is reported to have used standard tube oil colors from an art supply store and synthetic bristle brushes. He is quoted as saying, “I limited my palette to colors Haberle would have had (2).” XRF analysis of the present painting shows that Trotter was less exacting in selecting his paints for this work. Many anachronistic colorants were detected. Figure 6 shows the XRF analysis of the white paint used in the central white structure on the horizon (inset detail). Although one would expect lead white or perhaps ZnO to be used in the 1860s, the latter pigment being available at least since 1803 (10), the strongest peak in the XRF spectrum is in fact titanium. Trotter’s use of TiO2 white explains the transparency of the surface image in the radiograph in Figure 5. Based on this evidence alone, one could confidently rule out the signature date of 1866. Other colorants inferred from XRF data included Prussian blue [Fe4[Fe(CN)6]3•14-16H2O] in the blue window sills, yellow ochre in the yellow buildings and sunset [FeOOH + silicates], and red ochre in the central red building [Fe2O3 + silicates], all of which would have been available to 19th-century folk artists. 11 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 6. XRF spectrum with major element peaks identified for the white paint from the central white building. The inset shows a detail photo of the analysis spot.

In addition to these colorants, XRF revealed low levels of Co in all paints and the ground, Pb in varying amounts throughout the painting, and often concomitant peaks for Ba and Zn. The ubiquitous Co signal may represent a cobalt linoleate or naphthenate siccative added to accelerate the drying of the oil paints as confessed by Trotter. The relatively weak Pb signals in each spectrum are probably due to residual lead white ground left from stripping the reused canvas. When peaks associated with Ba and Zn occur together, this is often indicative of the use of lithopone, a co-precipitated mixture of ZnS and BaSO4 that has been used since 1874 as an inexpensive filler in many paints (10). Although XRF analysis provided potential pigments for most of the colors used in the painting, elemental spectra taken of the green hills showed only the omnipresent Co, Pb, Fe, Ba, and Zn. No metal could be clearly associated with a green pigment. To clarify the nature of the green colorant, the entire painting was placed under a gantry-mounted Raman microspectrometer, and vibrational spectral analysis was performed on the green pasture near the horse. The resulting Raman spectrum, shown in Figure 7(a), reveals numerous sharp spectral features indicative of an organic or organometallic pigment. The spectrum is compared to that of (b) phthalocyanine green, PG7, a chlorinated copper phthalocyanine complex first synthesized in 1938 (12), and a high degree of correlation is observed. However, under the microscope copious intermixed yellow particles were also visible, although they did not give as clear a Raman spectrum as the green component under the experimental conditions utilized. 12 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 7. Raman spectra of (a) green pigment particles and (b) reference phthalocyanine green, PG7.

Invasive Scientific Analysis Small loose fragments of the green paint analyzed above were extricated for analysis by transmission FTIR microspectroscopy. The green paint layer was separated from the other layers under the stereomicroscope prior to preparation for analysis. The resulting spectrum is shown in Figure 8(a) along with reference spectra of (b) polymerized linseed oil, and (c) hide glue. The sample spectrum shows a large oil component as revealed by methylene CH stretching bands at 2923 and 2852 cm-1 as well as the prominent νCO of the triglyceride esters at 1734 cm-1. The broad νOH peak centered at 3420 cm-1 indicates that the oil binder, presumably linseed oil, is well-cured and significantly hydrolyzed. Although Trotter did reveal that he used standard tube oil colors, he never mentioned the addition of a proteinaceous component, which can be presumed present due to the weak Amide I, II, and III bands at 1651, 1533, and 1450 cm-1. There are some verbal indications that Trotter did not always work in oils. At least one fake painting, an image of a ship at sea, is described by the Maine Antique Digest as being “tempera,” suggesting an egg binding medium, and Trotter himself in an interview with the Digest after sentencing cryptically described his first fake as being a “buttermilk paint,” presumable containing a casein binder (2). At this point the rationale for the protein in the green paint, whether intentional or accidental, could not be known. With no clear indicator of the pigments used in the green paint passage based on the FTIR spectrum in Figure 8(a), a spectral subtraction was attempted to remove the overwhelming spectral features of the binding media. After scaled 13 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

subtraction of the reference spectra for linseed oil and hide glue, the resulting residual spectrum is shown in Figure 9(a). Although this spectrum appears quite noisy, a spectral library search yielded a high quality match with the spectrum of Hansa Yellow, PY3, first available in 1928 (12). The spectrum of the pure colorant is shown in Figure 9(b). Because of its high tinting strength, this strongly colored yellow pigment is often mixed with a barium sulfate (BaSO4) filler, which is evidenced here by the sulfate stretching band triplet between 1235 and 1035 cm-1 and the sharp associated peak at 984 cm-1. The spectrum of pure barium sulfate is shown in Figure 9(c) for comparison.

Figure 8. FTIR spectrum of (a) green paint sample compared to reference spectra of (b) polymerized linseed oil and (c) hide glue.

Through a combination of FTIR and Raman analyses, the green paint used for the landscape appears to be a mixture of phthalocyanine green and Hansa yellow, which incidentally is often given the paint color name Permanent Green Hue in the modern artist’s palette (10). Permanent Green was originally a mixture of chrome oxide green with zinc yellow (ZnCrO4), which in fact would theoretically have been available to Haberle in the 1860s (10). It is possible that Trotter may not have been aware that the modern variant no longer uses the toxic chromate containing pigment. 14 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 9. FTIR residual spectrum of (a) green paint after subtraction of linseed oil and hide glue reference spectra. Comparison is made to reference spectra of (b) Hansa Yellow (PY3) pigment and (c) barium sulfate.

To assess the nature of the surface coating, a thin scraping was carefully removed from the area of the horse’s pasture without disturbing the underlying green paint layer. FTIR analysis (not shown) provided surprisingly an almost perfect match to well-aged shellac. Although Trotter had specifically mentioned copal as the coating of choice for his last forgery (2), the ‘Haberle,’ it would appear that an insect resin rather than a tree resin was used in this instance. The expected role of the shellac, which can be difficult to apply thinly and evenly by brush, in producing a convincing fake is not known. The detection of shellac required further investigation as it typically fluoresces bright orange when unbleached, unlike the cool, milky fluorescence observed in the UVA-induced visible fluorescence image in Figure 3(a).

15 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Further analysis of the binding media and varnish was performed using PyGC-MS of surface scrapings. The pyrogram of the TMAH-derivatized sample taken from the horse’s green pasture is shown in Figure 10(a). For comparison, pyrograms for (b) linseed oil and (c) very light shellac (unbleached) are included. The major marker peaks for various artists’ materials are identified in Table I (13, 14). Py-GC-MS confirms the results from FTIR analysis by revealing the simultaneous presence of oil, protein, the Hansa Yellow pigment, shellac, and trace amounts of pine resin (colophony). The shellac is most likely unbleached due to the lack of chlorinated marker compounds that have recently been reported to occur in shellac that has been decolorized by the addition of a chlorine bleach (15). Copal varnish, which was reportedly used in other Trotter fakes, is shown not to have been used in this work due to the absence of methyl sandaracopimarate, the marker compound for Manila copal, at detectable levels (16).

Figure 10. GC-MS pyrogram of TMAH-derivatized (a) green paint sample, (b) linseed oil, and (c) very light shellac.

The excessive craquelure of the paint in Village Scene provided numerous opportunities to prepare cross sections of paint passages without causing a noticeable lacuna in the painting’s surface. Although an invasive approach, cross section analysis is one of the only ways to explore the working methods of an artist. To understand the layered structure of this painting, selective areas were sampled and cross sections prepared for analysis by optical microscopy. 16 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Figure 11(a) shows a photomicrograph of one section taken from the hindquarters of the horse in the painting’s foreground. Figure 11(b) shows the same section under UV irradiation. The stratigraphy of the section revealed in the two images records the process by which Trotter created the fake. At the lowest level, a chunky, fragmented white ground layer with carbon inclusions shows the incomplete removal of paint from the reused canvas. This layer is best viewed in a previously published cross section (11), but can be faintly seen in the lowest area of Figure 11(a). On top of this are two thinner, homogeneous modern white ground layers applied by the forger to prepare the reused canvas for painting. The lower of these two is only partly preserved to the far right in the cross section shown here. A translucent layer seen in Figure 11(a) separates these two grounds, and it is shown to be highly luminescent in Figure 11(b). This blue fluorescence is typical of proteinaceous materials (17). On top of the modern grounds are two green paint layers, a dark green and an upper yellow-green (the Permanent Green Hue paint layer discussed above), again interleaved with a fluorescent material consistent with a proteinaceous layer. The paint of the brown horse overlays the landscape colors, again sandwiched between fluorescent layers, the topmost of which appears to be a mostly continuous blue fluorescent layer with occasional orange fluorescent components.

Figure 11. Visible light photomicrographs of a cross section sample from the horse’s hindquarters in (a) normal illumination and (b) UV irradiation. Lower right scale = 100 μm.

It is interesting to note in the cross section the numerous thin separations that exist in all of the oil paint and ground layers. It is believed that Trotter intentionally violated the painter’s “fat over lean” rule, using a fast drying medium like animal glue overtop of a slower drying medium like linseed oil (9). This type of construction inevitably leads to the oil paint being pulled apart into small islands by the rapid contraction of the surrounding protein layers, especially when heated, thus inducing nearly instantaneously the evenly random craquelure observed in Village Scene. 17 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Table I. Major pyrogram peaks, their identification, and associated artists’ material source of the TMAH derivatized compound Retention time (min)

Peak Identity

Origin

1.2-1.3

TMAH

derivatizing agent

3.04

pyrrole

protein

3.31

methyl methoxyacetate

oil

3.63

2-methoxyacetic acid, methyl ester

oil

4.22

N-N-dimethylglycine methyl ester

protein

4.76

styrene

protein

4.84

glycerol, trimethylester

oil

4.59

1,3-dimethoxy-2-propanol

oil

5.11

hexanoic acid, methyl ester

oil

6.09

heptanoic acid, methyl ester

oil

6.20

butanedioic acid, dimethyl ester

oil

6.96

octanoic acid, methyl ester

oil

7.52

3-methoxy-2,2’-bis(methoxymethyl)-1-propanol

oil

7.62

2-chloro-N-methylbenzamine

Hansa yellow

7.75

nonanoic acid, methyl ester

oil

8.65

heptanedioc acid, dimethyl ester

oil

8.68

8-methoxyoctanoic acid, dimethyl ester

oil

9.34

octanedioic acid, dimethyl ester

oil

9.46

dimethyl phthalate

plasticizer?

9.98

nonanedioic acid, dimethyl ester

oil

10.59

decanedioic acid, dimethyl ester

oil

11.01

tetradecanoic acid, methyl ester

oil

11.16

undecanoic acid, methyl ester

oil

12.09

hexadecanoic acid, methyl ester

oil

12.33

siloxane

column

12.76

unidentified, but occurs in reference

shellac

13.06

octadecanoic acid, methyl ester

oil

13.22

derivative of aleuritic acid

shellac

13.37

butylated hydroxytoluene (BHT)

antioxidant?

13.50

tetramethyl derivative of jalaric acid

shellac Continued on next page.

18 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

Table I. (Continued). Major pyrogram peaks, their identification, and associated artists’ material source of the TMAH derivatized compound Retention time (min)

Peak Identity

Origin

13.70

tetramethyl derivative of shellolic acid

shellac

13.97

derivative of aleuritic acid

shellac

14.19

dehydroabietic acid, methyl ester

pine resin, trace

14.21

derivative of aleuritic acid

shellac

14.34

siloxane

column

14.47

derivative of aleuritic acid

shellac

14.64

7-methoxy-tetrahydroabietic acid, methyl ester

pine resin, trace

15.18

7-oxo-dehydroabietic acid, methyl ester

pine resin, trace

15.21

7,15-dimethoxytetradehydroabietic acid, methyl ester

pine resin, trace

In numerous cross sections the penultimate surface layer was found to show traces of an inhomogeneous orange luminescent coating. This is consistent with the presence of unbleached shellac, which fluoresces a characteristic bright orange under UV excitation (15, 17). These cross sections confirm the FTIR and PyGC-MS analyses that also indicated shellac along with pine resin. The presence of the coating largely in the penultimate surface layer, rather than the uppermost one, explains the cool fluorescence in the UV-induced visible fluorescence image, Figure 3(a), rather than a warm orange fluorescence typical of unbleached shellac. A topmost surface coating of glue and dirt filters the UV radiation and prevents fluorescence from the largely underlying shellac. The presence of shellac as a picture varnish is unusual (17) aside from a few notable examples (15), although it may have been added here to the uppermost layers to induce hardness to the paint surface that could not easily be achieved in a young oil paint (9) or to enhance the craquelure through shrinkage. This combination of surface layers containing glue, shellac, and pine resin is surprisingly identical to the layering found in another fake, a purported 15th century portrait group acquired in 1923 by the National Gallery in London (18).

Conclusion A careful investigation of the artistic composition, materials, and construction techniques of one of Trotter’s forgeries, namely Village Scene with Horse and Honn & Company Factory, has revealed numerous ‘red flags’ indicating that the work is not a genuine piece of 19th-century folk art. Upon casual observation, the painting appears to have all the hallmarks of a great piece of primitive-style art. This is in fact one of the indicators of its ersatz nature – it has nearly all of the most prized physical characteristics of the folk art genre in one painting: a 19 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

quaint composition, naïve sense of perspective, female autograph, visible mattress ticking, heavy patina and fine craquelure suggestive of aged paintings, and mold stains commensurate with years of hanging on uninsulated parlor walls. Several curious and in some instances inexplicable clues are evident with merely a close visual examination. Foremost, the use of mattress ticking as a canvas lining fabric is unique in the authors’ experience, especially when no obvious canvas defects are visible to suggest relining was warranted. The applied nature of the mold and the uniform, dense surface cracking are also atypical of true primitive style paintings. Finally, the tacking margins reveal an implausible situation where the artist’s ground layer and painted composition appear to have been applied after the canvas was relined. Advanced imaging techniques also reveal several indicators of the painting’s speciousness. Again, for a relined canvas, there are no signs in the UV, NIR, or X-ray images of damages, losses, or structural deficits that would explain the lining fabric, which is shown to contain anachronistic optical brighteners. Furthermore, radiography revealed that no heavy metal pigments were used in this work, which is only feasible with a purely modern palette, although amorphous radio-opaque remnants of the original ground layer from a reused “stripped” canvas are detectable. In the event that scientific analysis is possible, a Trotter fake can be definitively identified as a 20th-century product due to the presence of numerous synthetic organic and inorganic pigments that were unavailable in the previous century or by the unconventional artistic technique of interleaving animal glue and paint as seen in fluorescence microscopy of cross sections. Since leaving prison, Robert Trotter has continued to paint 19th-century style artworks, but this time ‘genuine’ fakes that are sold legitimately to buyers of contemporary folk art (2, 3). When asked about the convicted forger’s new career, Arthur Riordan, one of the art dealers previously fooled by Trotter and owed compensation under the court sentencing, declared, “Good, I hope he makes a million dollars because we get the first $62,000 (2).” For the remaining thirty-nine unidentified Trotter fakes, compensation to their owners seems unlikely. Still, it is hoped that for the sake of art history and the reputations of gallery owners and collectors that the indicators revealed here might help to unmask others of Trotter’s oeuvre. The consistency of the forger’s methods is not at present known. However, four additional Trotter fakes confiscated by the FBI are now part of the Yale University Art Gallery’s study collection. Future work will hopefully subject these forgeries to the same level of scientific scrutiny in order to establish the reliability of the ‘red flags’ discovered here in the condition, construction, and materials of Village Scene.

Acknowledgments This work was completed at Buffalo State College and first appeared in a summary version as a presentation at the 2007 Annual Meeting of the American Institute for Conservation (11). The faculty recognizes the exhaustive research of the Trotter trial by graduate student Jennifer DiJoseph from the Class of 2010 and prosecutor Peter Jongbloed, former U.S. District Attorney in Connecticut, 20 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

for information regarding the case and ultimately the transfer of the Village Scene painting to Buffalo State College. GDS and CER acknowledge the financial support of the Andrew W. Mellon Foundation.

References 1.

Downloaded by 89.163.34.136 on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch001

2. 3. 4. 5. 6.

7. 8.

9. 10.

11. 12. 13. 14.

15. 16. 17. 18.

Jongbloed, P. S. United States v. Robert Lawrence Trotter, Criminal No. N89-59 (AHN); U.S. Department of Justice: District of Connecticut, 1990. Pennington, S. Maine Antique Digest. (March 1990), pp 14-A−17-A. Hewett, D. Maine Antique Digest. (December 1997), p 8-A. Grant, M. S. In Conserve O Gram, Number 1/9; National Park Service: Washington, DC, 2000, pp 1−3. Grant, M. S. In Conserve O Gram, Number 1/10; National Park Service: Washington, DC, 2000, pp 1−4. Mustalish, R. A. In Traditions and Innovation: Advances in Conservation, Contributions to the Melbourne Congress; Roy, A., Smith, P., Eds.; IIC: London, 2000, pp 133−136. Kushel, D. A. Stud. Conserv. 1985, 30, 1–10. Mecklenburg, M.; Lopez, L. F. In The Care of Painted Surfaces: Materials and Methods for Consolidation and Scientific Methods to Evaluate their Effectiveness; Il Prato: Padova, Italy, 2006, pp 49−58. Hebborn, E. The Art Forger’s Handbook; Overlook Press: Woodstock, NY, 1997, pp 148−152. Pigment Compendium: A Dictionary of Historical Pigments; Eastaugh, N., Walsh, V., Chaplin, T., Siddall, R., Eds.; Butterworth-Heineman: Oxford, England, 2004. Hamm, J.; Smith, G. D.; Kushel, D.; DiJoseph, J. AIC Paintings Specialty Group Postprints 2008, 20, 62–66. Lomax, S. Q.; Learner, T. J. Am. Inst. Conserv. 2006, 45, 107–125. Colombini, M. P.; Bonaduce, I.; Guatier, G. Chromatographia 2003, 58, 357–364. van den Berg, K. J.; Pastorova, I.; Spetter, L.; Boon, J. In ICOM Committee for Conservation, 11th Triennial Meeting, Edinburgh, Scotland; Bridgland, J., Ed.; James and James: London, 1996; pp 930−937. Sutherland, K. J. Inst. Conserv. 2010, 33, 129–145. Scalarone, D.; Lazzari, M.; Chiantore, O. J. Anal. Appl. Pyrolysis 2003, 68-69, 115–136. Eastaugh, N. The Picture Restorer; (Spring 2003), pp 11−12. Wieseman, M. E. A Closer Look: Deceptions and Discoveries; National Gallery Co.: London, 2010; pp 36−38.

21 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Chapter 2

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Scientific Examination and Treatment of a Painting by Gijsbert Gillisz d’Hondecoeter in the Mauritshuis Lauren Paul Bradley,*,1 Sabrina Meloni,*,2 Erich Stuart Uffelman,3 and Jennifer L. Mass4 1J.

Paul Getty Museum, 1200 Getty Center Drive, Suite 1000, Los Angeles, CA 90049-1687 2Royal Picture Gallery Mauritshuis, Mauritshuis, Korte Vijverberg 8, 2513 AB Den Haag, Postbus 536, 2501 CM Den Haag, The Netherlands 3Department of Chemistry, Washington and Lee University, Lexington, VA 24450 4Scientific Research and Analysis Laboratory, Conservation Department, Winterthur Museum, Winterthur, DE 19735 *E-mail: [email protected]; [email protected]

Gijsbert Gillisz d’Hondecoeter’s (1604-1653) panel painting, Cock and Hens in a Landscape, recently underwent complete treatment and technical examination at The Royal Picture Gallery Mauritshuis, The Hague (inv. no. 405). The interdisciplinary application of art historical research, conservation methodology, and scientific investigation led to several discoveries about the painting, including the revelation that major compositional elements of iconographical significance had been overpainted at some point in its history. Technical examination suggested that the original paint was in sufficiently good condition for the overpaint to be removed. The painting is currently on permanent display at the Prince William V Gallery in a state closer to the painter’s original artistic intent.

© 2012 American Chemical Society In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Introduction One of the themes of this book is to illustrate the cooperative efforts between scientific, academic, and museum communities in gaining new knowledge about cultural heritage material and using it to educate both the general public and students at the undergraduate and graduate levels. This chapter arose from a triangle of interactions between the Winterthur/University of Delaware Program in Art Conservation (WUDPAC), The Royal Picture Gallery Mauritshuis, and Washington and Lee University (W&L)—a partnership that has now been in place for over six years. During Lauren Bradley’s final year of study in the WUDPAC program, as the American Friends of the Mauritshuis Intern in Conservation, she undertook the treatment and technical study of Gijsbert Gillisz d’Hondecoeter’s Cock and Hens in a Landscape (collection ID MH405) under the direction of Mauritshuis Paintings Conservators Petria Noble and Sabrina Meloni (1, 2). Erich Uffelman, from W&L, assisted with the pXRF analysis performed on the painting and Jennifer Mass from the Winterthur Museum and WUDPAC carried out the SEM-EDS analysis. Both Uffelman and Mass participated in the writing of this chapter. The Mauritshuis has one of the world’s greatest collections of paintings from the Dutch Golden Age; WUDPAC has one of the world’s leading graduate programs in art conservation; and W&L has pioneering courses in using study abroad to educate science and non-science undergraduate students about the technical examination of 17th-century Dutch paintings (3, 4). This monograph is thus intended not only as a contribution to the art conservation literature, but also to be useful in various undergraduate courses on Chemistry in Art (5), as well as in the NSF Chemistry in Art Workshops (6). [Please also see chapters by Lang; Gaquere-Parker and Parker; and Hill in this volume.] Thus, we will briefly discuss the methodology behind the treatment and examination of Hondecoeter’s Cock and Hens in a Landscape, how the Hondecoeter family fits into the established history of 17th-century Dutch painting, the picture’s art historical context, the technical research findings, and the conservation treatment. This approach emphasizes the interdisciplinary linking of relevant cultural, socioeconomic, and historical information that undergirds the competent analysis and treatment of a painting.

Methodology Prior to embarking on a conservation treatment project, the painting or art object is thoroughly examined and documented, adhering to the guidelines and standards of ethical practice established by the field (7). Because Hondecoeter’s Cock and Hens in a Landscape had a complex surface that raised questions about the condition and the restoration history with implications for treatment, a variety of analytical techniques were used to study the painting on both a macroscopic and a microscopic level, including magnification under a binocular microscope, ultraviolet-induced visible fluorescence imaging (UV), infrared photography (IR), X-radiography, handheld portable X-ray fluorescence spectroscopy (pXRF), cross-sectional microscopy with UV and visible illumination, polarized light microscopy (PLM), Fourier transform infrared spectroscopy (FTIR), scanning 24 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS), and SEM in the backscattered electron mode (SEM-BSE). A very well-illustrated text that cogently summarizes these conservation science techniques has recently appeared (8). To supplement the technical data, a number of related works by Hondecoeter and his contemporaries were examined in other collections including the Rijksmuseum in Amsterdam, the Museum Boijmans van Beuningen in Rotterdam, and the Philadelphia Museum of Art in Pennsylvania, USA, in addition to several pictures at Pieter de Boer’s Gallery in Amsterdam. These paintings provided insight into Hondecoeter’s painting practices in addition to establishing a visual reference for how a Hondecoeter panel in good condition should appear today. Examination generally begins by studying a painting using non-destructive imaging techniques such as magnification under visible light, UV, IR, and X-radiography. Magnification is an invaluable tool for the conservator; a trained eye can draw sophisticated conclusions about a painting’s condition and the artist’s working practice using magnification alone. UV irradiation provides additional information about the surface and is often used for identifying areas of restoration/overpaint as newer materials typically fluoresce differently than the aged, original material. For example, aged natural resin varnishes will fluoresce green under UV irradiation, while synthetic polymer-based materials such as acrylic paint (often used in conservation) will appear dark. Fluorescence colors can also aid in assigning preliminary pigment identifications. IR can be a useful technique for studying aspects of the picture hidden from view beneath layers of paint such as an artist’s preparatory sketch or pentimenti (changes to the composition). If there is enough contrast between, for example, the lines in the artist’s sketch and the surrounding ground, and if the proper wavelengths are selected, the intervening paint layers can become IR transparent. The IR radiation penetrates through to the ground layer and is reflected back through the paint layers to the detector, making it possible to image the hidden lines. If the ground layer is dark in color, less IR radiation is reflected, which decreases the ability to distinguish underdrawings from the surrounding ground (9, 10). The success of X-radiography in imaging a painting is based on differences in radio opacity of different artists’ materials. The technique is useful for visualizing structural components such as panel or canvas joins, and the painting technique, especially when pigments containing heavy metals such as lead white (2PbCO3.Pb(OH)2) or vermilion (HgS) are present (11). Scientific instrumentation can be used to address questions that are not possible to answer using imaging techniques alone. For example, pXRF provides information about the elemental composition of a painting’s surface, which can be used to infer which pigments may be present. X-ray radiation has sufficient energy to penetrate the entire painting structure, meaning the spectra will typically contain information about numerous paint layers including the ground layer. When coupled with the use of a vacuum pump, pXRF can identify elements with atomic numbers of 13 (Al) or greater. It is an excellent technique for analyzing works of art because it is non-invasive and non-destructive and the data collection is relatively fast; it is often possible to get discriminatory data in two minutes or less. Limitations include the inability to target a specific 25 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

layer in the painting structure and insensitivity to elements with a lower atomic number for air path analysis. Overlapping X-ray lines and the presence of elements that could be associated with multiple compounds may also complicate qualitative interpretation of the spectra. pXRF can be a powerful tool for pigment characterization, especially when used in combination with other elemental analysis techniques or with imaging and molecular analysis methods such as cross-sectional microscopy, polarized light microscopy, SEM analysis, and FTIR. Excellent introductory discussions of the strengths and limitations of pXRF spectroscopy have recently appeared (12, 13), and a book on the applications of handheld pXRF in art and archaeology is in press (14). Techniques which require sampling may be used when further questions arise about a pigment or a specific layer in the painting structure. Samples are typically taken under magnification from an area of the painting with pre-existing damage or at the painting’s edge, where the surface is hidden from view by the frame rabbet (15). Depending on the sample size required for analysis, a tungsten needle (16) or a scalpel can be used to take the sample, which may range from a few microns in diameter to the size of a period at the end of a 12-pt font sentence. Cross-sectional microscopy involves polishing a sample embedded in resin to reveal the sample’s edge, providing a cross-sectional view of the layer structure. The polished sample can be examined under magnification using light microscopy and/or scanning electron microscopy. Cross-section samples provide information about the stratigraphy of paint and varnish layers; illumination under the microscope using a UV irradiation source coupled with different wavelength filter cubes can help clarify the boundaries between layers or the presence of varnish and glaze layers. SEM-EDS can be used to identify the elemental composition of a single pigment particle within a cross-section sample, to identify the presence of pigment alteration/degradation products, or to create elemental X-ray maps of the entire sample, providing detailed information about the distribution of elements throughout the layering structure. This technique complements pXRF data in that it allows for the detection of elements as light as boron (Z = 5) because it is performed under high vacuum. SEM-BSE produces grayscale images based on the local average of the atomic numbers present and is invaluable for studying the particle morphology of paint pigments and fillers (17). FTIR is a powerful technique for characterizing general classes of organic materials such as oils, polysaccharides, proteins, resins, and waxes, as well as for identifying synthetic resins, inorganic pigments (particularly polyoxoanions), and natural minerals. In the transmission mode, the technique involves taking a microgram-sized sample, flattening the sample on a diamond half cell using a steel micro-roller, and measuring the IR absorption across the spectral range of 4000 to 650 cm-1 (18, 19).

Hondecoeter Family Gijsbert Gillisz d’Hondecoeter (1604-1653) came from a family of artists; his grandfather (Nicolaes Jansz), his father (Gillis Claesz), his brother (Nicolaes II), and his son (Melchior) (20) were all painters (21, 22). The tradition of passing 26 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

craft-based knowledge down from generation to generation was widespread in the 17th century, when painters learned through the apprentice system regulated by the St. Luke’s Guild. Many young painters trained under a family member or a familial connection made through marriage; others held apprenticeships with a master for a fee. Relatively little has been written about Gijsbert Hondecoeter or the Hondecoeter family. Gijsbert trained under his father, Gillis Claesz, who predominantly painted landscapes populated by animals. Gijsbert was primarily active in Utrecht, registering with the city’s St. Luke’s Guild in 1626 (23). His pictures, which primarily depict live game birds and poultry, are fairly static with little drama or interaction between the birds. Gijsbert trained his son, Melchior, until his death in 1653. According to Arnold Houbraken, an 18th-century biographer of 17th-century Dutch artists, Melchior continued painting under his uncle, Jan Baptist Weenix, who was married to his father’s sister, Justina. It is likely that Melchior worked alongside his cousin, Jan Weenix, in his uncle’s studio (22). Melchior remains the best-known member of the Hondecoeter family and is perhaps the greatest exponent of the poultry-yard genre. His paintings are characterized by lively compositions, accuracy of anatomical description, and expressive use of color. In comparison to the rest of 17th-century Europe, the Dutch art market was unique. Following the Protestant Reformation in the 16th century, there was a sharp decline in ecclesiastical art patronage throughout the Northern Netherlands. Because Calvinist theology prohibited the use of images for worship, few Dutch painters depicted Biblical or devotional scenes (24, 25). Furthermore, there was little court patronage as the Dutch aristocracy was relatively small and the government was dominated by a regent class. Dutch business and trade were highly prosperous, leading to the formation of what might be considered the first modern economy and to the formation of a substantial middle and upper class in addition to a wealthy elite (26–30). A new and thriving art market emerged to satisfy the demand for pictures created by the affluent members of the Reformed Protestant Dutch Republic. Dutch artists produced millions of paintings during the 17th century. Although some of these pictures were commissioned works, a substantial majority was prepared for the open market, which was typically regulated at the local level by the St. Luke’s Guild of each city or town. Rembrandt van Rijn, arguably the best known painter from the Dutch Golden Age (31), was unusual in that his pictures could not be grouped within a single category—he painted religious and historical paintings in addition to portraits, landscapes, still-lifes, and genre scenes. An overwhelming majority of Dutch painters specialized in a particular category or sub-category of painting in order to produce paintings more efficiently and to minimize their competition. Thus, an artist might not only specialize in still-life painting, but further specialize in breakfast still-lifes, “ontbijtes,” or luxury still-lifes, “Pronkstilleven” (32, 33). Patrons of pictures depicting domesticated and wild birds were likely rich burghers who kept exotic and native fowl on their country estates. Occasionally, poultry breeders would commission artists to paint prize birds and common farmyard specimens rather than purchasing already completed works (34). 27 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

The Painting Hondecoeter’s Cock and Hens in a Landscape (Figure 1) has a complex surface that was not fully understood until after treatment began. The picture depicts a large yellow cock and two hens in an outdoor setting. The cock sits on an overturned woven basket at the center of the composition while a brown hen balances on the rim of another basket at left and a black hen rests at right. An atmospheric landscape recedes into the distance at right, behind a potted plant and a terracotta vessel. Prior to treatment, the lower right corner also contained a leafy green plant, which was determined to be a much later addition, as the plant covered small losses and cracks in the original paint layer (revealed by examination with the binocular microscope). The decision was made to remove the non-original plant during cleaning, which revealed a previously unknown animal skull, a bone, and a black and white rabbit (Figures 2 and 3). Black and white rabbits are a fairly common motif within Hondecoeter’s oeuvre, however, the depiction of a skull and bone seems to be unique to the Mauritshuis painting. The plant was most likely added to hide the fact that the rabbit’s body was truncated when the panel was cut down, and perhaps also because skulls, as memento mori, or reminders of death, had lost their appeal to later audiences. Vanitas symbols featured frequently in 17th-century Dutch painting, reflecting the conservative Calvinist religious views of the time (21, 24). The plant was added before the painting entered the Mauritshuis collection in 1876, as the rabbit and skull are not visible in early photographs, nor are they referenced in early descriptions. The painting has a gray-blue sky, which was repainted at some point to be brighter in color. Hondecoeter repeated motifs and compositional arrangements throughout his career. This practice lent itself to greater productivity, as less time was spent innovating during the planning stages. The large woven basket, the yellow cock with a spotted breast, the downward facing hen, the black and white rabbit, and ribbon-like strands of grass all appear in other works by Hondecoeter, suggesting the Mauritshuis painting may have been created for the open market rather than for a specific commission. The brushwork used to render these forms is similar to representations found in pictures examined at Pieter de Boer’s gallery, suggesting Hondecoeter was well practiced at their depiction. The Mauritshuis composition, with a landscape receding in the distance and several large birds in the foreground, is almost identical to works attributed to Hondecoeter’s Dordrecht-based contemporary, Aelbert Cuyp (1620–1691). A 1935 Mauritshuis catalogue links the two artists, referencing a painting by Cuyp sold at auction, which had a similar composition and a boat in the background (35, 36). Unfortunately, no further information is provided, leaving many unanswered questions about the picture’s provenance, its attribution, and its whereabouts today. A reproduction of a painting that matches this description was found in the photographic archives at the Netherlands Institute for Art History in The Hague (RKD); further research would be needed, however, to draw any conclusions.

28 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Figure 1. Gijsbert Gillisz d’Hondecoeter (1604-1653), Cock and Hens in a Landscape, no date. Oil on panel, 52 x 70 cm. Royal Picture Gallery Mauritshuis. Inv. no. 405. Before treatment (2010). Courtesy of The Royal Picture Gallery Mauritshuis. (see color insert)

It is possible that Cuyp was influenced by Hondecoeter’s work or that both artists were working from a similar model in a workshop book. [Two useful introductions to Dutch workshops have been published (37, 38).] Archival research carried out at the RKD indicated that historically, there has been confusion over the attribution of poultry-yard scenes by Hondeoceter and by Cuyp. Another picture attributed to Hondecoeter at the Mauritshuis was previously attributed to Cuyp and has a false Cuyp signature, added sometime before the painting entered the collection in 1899. Examining works by Hondecoeter in other collections provided insight into how he typically painted the sky and how he positioned his birds in relation to the edges of the composition. Studying these features was significant because the Mauritshuis panel has been cut down and the original sky was repainted. Hondecoeter’s Waterfowl (A 1332), dated 1652, at the Rijksmuseum Amsterdam, served as a reference for how a Hondecoeter panel in good condition should look today. The upper half of the painting is dominated by a gray-blue sky with several small clouds; dark, horizontal striations resulting from the formation of lead soaps in an underlayer are visible throughout the sky. [Lead soaps are discussed in more detail in the Condition Section.]

29 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Figure 2. Detail, lower right corner during cleaning. The varnish and overpaint are intact at left and have been removed at right. (see color insert)

Figure 3. Cock and Hens in a Landscape. Royal Picture Gallery Mauritshuis. After treatment (2011). (see color insert) 30 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Construction The painting was executed on an oak wood panel comprised of two glue-joined horizontal planks with a radial grain direction (Figure 4). The panel verso has no evidence of original beveling, which would have facilitated fitting the painting into a frame (39). Shallow grooves, approximately 1 cm in length run perpendicular to the panel edge at the top and bottom of the right side as seen from the reverse (Figure 5). These grooves appear to be part of the original construction and may have served an analogous function to beveling. Similar marks have been found on Rembrandt van Rijn’s Supper at Emmaus from the Musée du Louvre (INV#1739). On Rembrandt’s painting, which has been thinned and cradled, the verso has been described as having traces of “shallow gouged grooves as was sometimes done instead of beveling” (40).

Figure 4. Panel verso showing the outline of the secondary support that was once attached; the joint runs horizontally through the center of the panel.

In preparation for painting, a thin layer of chalk (likely glue-bound CaCO3) was applied to the surface, filling the open pores and interstices of the wood grain. This practice not only provided the artist with a smoother surface on which to paint, but also reduced costs, as chalk was a less expensive material than pigment bound in drying oil.

31 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Figure 5. Detail, shallow grooves in upper right corner of the panel verso. Raking light from above.

The chalk layer was followed by a gray imprimatura underlayer. The imprimatura creates a uniformly colored surface for painting in addition to setting the overall tonality of the picture; i.e., despite often being concealed under several layers of paint, the imprimatura affects the overall appearance of the painting (41). Dark imprimatura layers often remain visible on the surface, functioning as the midtone or the shadow (42). This was recently observed, for instance in a Teniers/Brueghel study (43).

Figure 6. Cross-section sample (405 x01) taken from the overpainted sky, 400x magnification; reflected light at left, ultraviolet irradiation at right. (see color insert) 32 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Cross-sectional microscopy coupled with SEM-EDS analysis indicates the imprimatura layer contains a mixture of lead white, an organic black pigment, and a calcium-based filler material. Red particles are visible near the bottom of the imprimatura layer under 400x magnification in reflected light (Figure 6). These particles were identified as Pb3O4 using SEM-EDS and may be a byproduct of lead soap formation rather than an intentional additive to the paint. Because the imprimatura has a lead white matrix, the brushstrokes used to apply the layer are visible in the X-radiograph. The strokes are broad and intersect in large crisscross patterns. Examination with IR (Artist multispectral imaging camera, IR2 mode, with a long wave pass filter 1000 nm) revealed evidence of underdrawing lines in the cock that are not visible under normal lighting conditions. The lines have a slightly broken quality, suggesting they were created using a dry medium such as black chalk or charcoal rather than with a wet medium and a brush. Brush-applied drawings have a more fluid, unbroken line quality. A distinct underdrawing line runs along the back of the cock’s head, along the upper edge of his comb, and around his wattle. The location of the preparatory marks visible suggests Hondecoeter used underdrawing to plot the location of the primary compositional forms, but did not go so far as to create shadow or any indication of texture. The paint application technique used to create the picture was studied using a binocular microscope (10x – 50x magnification) and cross-sectional microscopy (40x – 400x magnification). Examination indicated the composition was worked up by painting the largest forms first, namely the birds and the central basket. The landscape was filled in around the birds, working from the background to the foreground. There are no major pentimenti or changes to the composition, further confirming the placement of the figures and forms had already been established when painting began. The blue sky was applied at the beginning of the painting process, directly on top of the gray imprimatura, without leaving reserves for the birds or the landscape. Leaving an area in reserve is an artistic technique in which other parts of the composition are painted around an area, which the artist will later fill (42). This approach minimizes the amount of paint used and reduces the likelihood that an underlying form will become visible over time as the upper paint layers become increasingly transparent. As paints age and oxidize, the refractive index of the binding medium increases, becoming closer to that of the pigments, which causes the paint layer to become more translucent; lead soap formation also increases translucency. The paint used to render the sky was found to contain a mixture of lead white and smalt (a cobalt blue glass). [vide infra] The large, woven basket and the birds were worked from dark to light with the highlights added last. Details such as the black markings on the brown hen, the spots on the cock’s breast, the eyes, the beaks, and the claws were also added towards the end of the painting process. Dry brushstrokes were used around the edges of the birds to soften the edges of their forms and to make them sit more convincingly in space. The painting has a relatively smooth surface with little impasto texture. The paint used to create the landscape along the horizon line and the shadowed side of 33 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

the basket in the center is thin and medium-rich. Washy brushstrokes are visible on the shadowed side of the large basket. Hondecoeter used glazes in certain areas to achieve a richer surface appearance. Glazing involves the use of a thin, transparent paint to modulate or enrich the colors in the underlying paint layer. This technique is apparent in the black hen’s comb and in the landscape. On the hen’s comb, the glaze is visible under magnification as a fractured, translucent red layer on top of a more opaque red paint. The glaze has a bright pink fluorescence in UV, which is characteristic of an organic red pigment such as red lake. pXRF analysis suggested the opaque red paint contains vermilion (HgS). In the landscape, there is a non-fluorescent, translucent brown layer pooled within the interstices of the underlying paint texture. This layer likely represents the combination of an original copper-containing glaze that has discolored over time and the remains of an old, discolored varnish. The presence of copper was confirmed using pXRF. [A more extensive discussion of pXRF and paintings analysis may be found in the following chapter (44).]

Condition and Previous Treatment History Prior to treatment, the painting’s condition made it unsuitable for display. There were multiple layers of thick, discolored varnish on the surface, which distorted the tonal relationships and made it difficult to interpret the darker passages (45, 46). The varnish had an uneven surface gloss with the darker passages becoming matte and crazed (micro-cracking). The Mauritshuis records describe several restorations dating back to 1916, when Mauritshuis restorer Derex de Wild performed the first documented treatment. De Wild reduced a thick layer of varnish on the painting “by half” using a mechanical technique (47); this may have entailed rubbing the surface of the degraded, brittle varnish until it flaked and turned into a powder. The thinned varnish was regenerated several times and a new varnish was applied. Regeneration frequently involved the use of alcohol vapors and copaiba balsam. De Wild’s report describes leaving the overpainted sky intact. Mauritshuis restorer J. C. Traas treated the painting again in 1937 (48). Traas performed structural work on the panel in addition to removing discolored varnish and overpaint from the surface. He repaired the joint between the two planks and although it is not explicitly referenced in the report, he probably also inserted the butterfly cleat spanning the joint on the verso. Traas applied a new varnish and retouched the painting. In early 2010, the painting was surface cleaned using a 1% solution of triammonium citrate in demineralized water. A saturating layer of 10% Regalrez 1094 varnish in Terpentina D was applied on top of the old varnish using cotton wool wrapped in silk. This was done in an attempt to resaturate the matte areas and make the painting aesthetically acceptable for display. This minimal treatment was unsuccessful and did not achieve adequate saturation (49–53). Further testing with Laropal A81 varnish and different concentrations of Regalrez 1094 varnish 34 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

also proved unsuccessful. At this point, it was decided that the painting would benefit from a complete treatment involving varnish removal. During the present examination and treatment campaign, it became clear that much of the painting’s treatment history went undocumented and likely occurred before the picture entered the collection in 1876. A variety of analytical techniques were used to investigate the history and to identify passages of disfiguring repaint that should be removed. Perhaps the most drastic interventions to the work were those made to the original format (the size of the support). The painting currently measures 52 x 70 cm, although the original dimensions are unknown. At some point, the panel was cut down on the right side, slightly trimmed along the upper edge, and sanded along the joint (39, 54). In the past, paintings were frequently altered to suit the needs of the collector; the picture may have been cut down to fit a frame or to serve as a pendant painting to a smaller piece. The major alterations at the right and top edges occurred before 1876, when the painting’s present dimensions first appear in a Mauritshuis inventory book recording new acquisitions to the collection (55). The sanding along the joint may have occurred as late as 1937 when Traas describes putting the joint between the two planks together; sanding was commonly done to achieve a better glue bond (48). The cock’s feathers are slightly discontinuous across the joint, indicating some of the image material was removed when the joint was repaired.

Figure 7. Diagram speculating how much of the original panel is missing at the top and left edges.

35 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

The back of the panel has a ghost image outlining where a non-original secondary support structure was once attached (Figure 4). The outline is truncated at left, indicating it was in place and removed before the panel was cut down. Based on the approximate symmetry of the secondary support’s vertical members to one another, it is possible that 21 cm are missing at the left (Figure 7). This calculation is based on the average distance between the vertical members, the average width of each individual member, and the distance between the member at the far right to the edge of the panel. The lower plank is approximately 5 cm wider than the upper plank, which may suggest as much as 5 cm are missing from the top. It was, however, not uncommon for panel planks in the 17th century to be asymmetrical in width. These approximations suggest the original dimensions were in the range of 52-57 cm in height and 91 cm in width, which is consistent with established height to width ratios of 17th-century marine-format panels (56). In the 17th century, artists could purchase supports according to standardized sizes intended for different subjects including marine scenes, landscapes, and portraits (57). It is significant to note that related poultry-yard scenes attributed to Cuyp, which are reproduced in the RKD archives, also appear to have a marine format. Prior to treatment, the painting had several layers of discolored yellow varnish on the surface. In addition, Traas had intentionally tinted the uppermost varnish layer using small black pigment particles when he applied it in 1937. Historically, tinting a new varnish was done to imitate the appearance of an aged yellow varnish, which was a valued aesthetic component of old master paintings in the early part of the 20th century—almost all of the paintings Traas treated at the Mauritshuis have a tinted varnish (58). This practice is no longer in use as it does not do justice to the artist’s original intention. During cleaning, after the varnish layers were removed, it became apparent that the original sky had been repainted using a bright blue paint, and several clouds were added at the center of the picture, above the birds. The overpaint had a fractured appearance under magnification along the top and right edges, indicating it was applied sometime before the panel was cut down. A second overpaint campaign, carried out using a slightly darker blue, was present along the joint (Figure 3). The overpaint was analyzed using a variety of noninvasive and microsampling techniques including pXRF, PLM, cross-sectional microscopy, FTIR, and SEM-EDS, in an attempt to date its application and to better understand how it might be removed. Only the earlier, broadly applied overpaint will be discussed here. Analysis performed using FTIR indicated that the bright blue overpaint applied across the entire sky contained a mixture of lead white (2PbCO3.Pb(OH)2) and Prussian blue (ferric ferrocyanide, Fe4[Fe(CN)6]3) bound in linseed oil. Prussian blue was synthesized in 1704 and first used as an artist’s material shortly thereafter (59–61), meaning the overpaint must have been applied sometime between that date and 1876, when the painting’s present dimensions first appear in print. FTIR can be used as a diagnostic indicator for Prussian blue because the pigment contains ferric ferrocyanide components that have a strong carbon nitrogen triple bond absorbance at 2083 cm-1, a region of the IR spectrum in which very few other artists’ materials are active. 36 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

The Prussian blue particles are barely perceptible in cross-section under 400x magnification, causing the layer of overpaint to appear white. This phenomenon can be attributed to the pigment’s incredibly fine particle size and its high tinting strength. Only a small quantity of this high tinting strength pigment is required to tint lead white paint blue (59). Prussian blue particles are approximately 0.01 to 0.02 µm in diameter and are best visualized using PLM, which involves dissolving the binding media, separating the pigments from the paint matrix, and examining them using transmitted light (59). These properties make it difficult to detect Prussian blue on a painting using pXRF alone, since iron is also a major component of surface dirt and earth pigments. Cross-sectional microscopy performed on several samples taken from the sky confirmed the presence of an uneven varnish residue between the overpaint and the original paint. This information had implications for the treatment, as oil-based overpaint can be difficult to remove from an oil painting due to similarities in solubility. Furthermore, lead-based oil paints have siccative properties—the lead ions act as driers, promoting oxidation within the film and the formation of crosslinks, resulting in a tough paint layer. Having a varnish underneath the overpaint is beneficial during cleaning because the varnish can be used as a sacrificial layer; cleaning solvents and gel formulations can be tailored to target the varnish rather than the overpaint, allowing the overpaint to be removed without resorting to harsh cleaning solutions. Hondecoeter’s original sky obscured by the overpaint was found to contain a mixture of lead white and smalt using cross-sectional microscopy and pXRF. Smalt is a pigment created by grinding potassium glass colored with blue cobalt oxide into a workable powder (K, Al, Co silicate); coarse grinds result in a bluer pigment, while finer grinds result in a grayer one (62, 63). Loose smalt particles have a characteristic conchoidal fracture pattern visible under high magnification. When embedded in a paint matrix and viewed in cross-section, the pigment often has a triangular shape as was observed in the Hondecoeter sample (Figure 6). In the 17th century, smalt was produced and sold in quality grades ranging from gray to blue. Compared to the costly pigment ultramarine blue, which was often reserved for commissioned paintings or passages of the painting with iconographical significance, smalt was a fairly inexpensive alternative (63). Characteristic peaks for cobalt and silicon (observed as a shoulder on the lead peak)—two principal elemental components of smalt—are present in pXRF spectra collected throughout the sky, corroborating the visual evidence of smalt observed in cross-section (Figure 8). Strong peaks for lead can be attributed to the presence of lead white in the overpaint, in the original sky, and in the imprimatura layer. Many varieties of blue smalt are inherently prone to discoloration over time (likely due to the pigment’s potash to silica ratio) (64–66), while some varieties are more stable and have retained their blue color over hundreds of years. Determining whether Hondecoeter’s original sky was in good condition played a major role in the decision whether or not to reveal it through cleaning. If the original sky was in a badly deteriorated state, the overpaint might have been left intact because there would not have been a significant visual gain through removing it. Cleaning tests confirmed the original paint surface had minimal evidence of abrasion. The 37 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

smalt particles appeared blue in cross-section and on the surface in areas where the overpaint was abraded. X-ray maps created using SEM-EDS (17) did not reveal any evidence of chemical degradation (migration of potassium ions outside of the smalt particles), further indicating that the smalt particles were in good condition (Figure 9). The maps show the potassium is still closely associated with the particle’s silica core. [Smalt degradation is discussed at greater length in the following chapter of this book (44).]

Figure 8. pXRF spectrum collected from an area of overpainted sky with characteristic peaks for smalt (Co, Si) [Bruker Tracer III-SD Rh X-ray tube at 15 kV and 55 μamps, 20 Torr, 1080 s].

Examination under the binocular microscope confirmed the plant in the lower right corner was a later addition. The overpaint used to render the plant had a coarse, pebbly texture with large white particles visible at the surface; this contrasted with the smoother surface of the finely ground original paint. The overpaint extended across age cracks in the underlying original paint and went over the right edge of the panel, where the painting had been cut down. Cleaning tests indicated the overpaint was readily soluble in the same solvents used to remove the varnish layers (vide infra). Lead soaps are forming preferentially along the wood grain in the gray imprimatura underlayer, resulting in the appearance of short, dark horizontal lines throughout the sky (67) (Figure 10). In other areas, lead soap aggregates have become mobile and migrated through adjacent layers, causing a deformation or rupture at the surface.

38 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Figure 9. Left: SEM-EDS X-ray map of potassium (purple) and silicon (blue) within cross-section sample 405 x01 -- illustrating the absence of smalt degradation (no migration of K+ away from the Si core). Right: Same sample, back scatter electron image showing the smalt particles (darker gray/black) within the cross-section (uncoated sample analyzed using variable pressure mode). (see color insert) Lead soaps have been studied extensively in the previous decade because they occur almost ubiquitously in old master oil paintings (68–70). They likely result from the slow base-induced ester saponification of paint layers containing lead white (or other lead-containing pigments or driers). [Oil paints containing zinc and copper pigments or driers, or smalt, are also prone to metal soap formation.] Oil paint contains fatty acid triglycerides and lead white pigment is a basic lead carbonate (2PbCO3.Pb(OH)2). Over time, the basic lead white cleaves the ester linkages formed between the triglycerides to produce glycerol and lead carboxylates, commonly referred to as “lead soaps”. In some cases, these lead soaps aggregate and form globules in the paint layer that can grow large enough in size to push through the surface of the paint layers, appearing as raised white bumps on the painting’s surface (this has not happened in this painting). That lead soaps can form preferentially along the wood grain is because the chalk ground in the interstices of the wood acts as a reservoir of free fatty acids that reacts with the lead white in the overlying lead-white containing imprimatura. This phenomenon has been studied in detail in several other paintings in the Mauritshuis (67). Due to saponification involving the lead white particles, the paint loses its opacity in these areas and becomes more transparent, appearing darker. This is occurring throughout the picture but is most noticeable in areas where the image paint on top of the imprimatura is abraded or has also become transparent—most notably in the sky and the yellow cock’s body. The lead soaps are visible in cross-section as translucent areas. X-ray maps created using SEM-EDS show a lower concentration of elemental lead in these areas, which can be attributed to a dissolution of the lead white pigment particles during lead soap formation. This phenomenon is also apparent in images created using SEM-BSE; the lead soap aggregates appear darker than the surrounding lead white paint matrix (Figure 11). Detecting differences in average atomic number is possible because the SEM-BSE intensity (the number of beam electrons elastically 39 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

scattered back towards the electron beam) is related to the atomic number of the atoms in the interaction volume of the electron beam (17).

Figure 10. Detail, before treatment, showing disturbing dark lines visible in the sky resulting from lead soap formation in the underlying imprimatura layer that correspond to the ground-filled wood grain.

Lead soap aggregates are larger in size than lead white pigment particles; accordingly, the formation of lead soaps in the imprimatura layer (corresponding to the chalk ground-filled interstices of the wood grain) has led to an increase in volume. The surface paint in these areas has plastically deformed as the result of this increase, whereas in other areas the paint has fractured and cracked. The expanded paint has created prominent ridges across the surface of the entire painting (71). This texture became especially apparent after the thick layers of varnish and overpaint were removed (Figure 12).

Treatment Treating the painting (54) was a complex endeavor due to the presence of multiple varnish and overpaint campaigns with varying solubilities. The extent of non-original material present was not fully understood until after cleaning began, as the thick discolored varnish layers made it difficult to interpret the underlying paint layers. The aged varnish had a strong fluorescence under UV illumination that masked the differences in fluorescence between the original and non-original material. More than half a dozen solvent mixtures and gel formulations were used during cleaning (72–77). 40 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Small cleaning tests were performed at the edges of the painting to establish which solvents or gel formulations could remove the varnish and later repaint without harming the underlying original paint. Testing continued throughout the treatment as new passages of overpaint became apparent. Frequent discussion among the Mauritshuis conservators and curators guided the decisions to proceed with each step of the cleaning process.

Figure 11. 1150× magnification. SEM-BSE image (uncoated sample analyzed using variable pressure mode) of a lead soap aggregate in the imprimatura layer; the aggregate is the large dark oval-shaped area at the center of the image (Sample 405 x01). Cleaning proceeded cautiously by removing one layer of non-original material at a time. The uppermost layers of discolored varnish were removed first using free solvent mixtures of isopropanol, isooctane, and acetone; these mixtures are consistent with what one would expect to use for dissolving an aged natural resin. During cleaning, it became apparent that the painting was selectively cleaned in the past; an even older natural resin varnish was present in the dark passages on the lower half of the painting beneath the upper, pigmented varnish. Darker colors are more vulnerable during cleaning due to a higher proportion of medium to pigment in the paint and a lack of good metal ions to promote extensive oxidation and cross-linking within the paint film as it dries (78). Furthermore, dark passages are often less satisfying to clean because the resulting visual change is not as great as it is with lighter passages. It is not uncommon to find partially removed coatings on paintings that have been cleaned in the past; this phenomenon is often referred 41 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

to as “port-hole cleaning” or “dealer cleaning”. Because this underlying varnish was older and more oxidized, it required a slightly more polar solvent mixture to remove. Varnish oxidation is an autoxidation process in which the terpenoids of the resin undergo various reactions with atmospheric molecular oxygen (45, 46).

Figure 12. Detail after cleaning and varnish removal, lower right corner. Raking light from the top showing the raised horizontal ridges created by lead soap formation in the imprimatura layer.

The restorer-applied plant in the lower right corner was removed because it covered a significant portion of the original composition and cleaning tests indicated the underlying original paint was in good condition. The overpaint was soluble in the same free solvent mixtures used to remove the varnish. In order to minimize mechanical action (movement of the cotton swab) on the surface, isopropanol gelled with 5% Klucel-G by weight was used to remove the plant. Klucel-G is a hydroxypropylcellulose that is soluble in water and alcohols. It was selected as a gelling agent because it was assumed to have no independent cleaning properties and could be used to increase the viscosity of the isopropanol solvent and decrease the evaporation rate, thus minimizing the amount of solvent introduced to the surface. Based on the solubility properties of the plant, the paint was likely a drained oil mixed with a natural resin varnish; this combination of materials was a fairly common restoration technique in the past. Absorbing or “draining” the excess oil binding medium from an oil paint before it is used results in a “leaner” paint that remains more soluble than full-bodied oils over time. 42 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

The blue repaint in the sky flattened the pictorial space and the bright color was inconsistent with what one would expect to find in a 17th-century Dutch picture. Testing indicated the repaint was insoluble in most free solvent mixtures appropriate for use on an aged oil painting. Mechanical removal of the repaint using a scalpel under the microscope would prove to be time consuming and hazardous due to the lead white component of the paint; furthermore, the coarse surface texture of the original sky would make it difficult to use a scalpel without harming the original surface. Further testing with benzyl alcohol and Pemulen TR2 indicated that a 5% benzyl alcohol Pemulen gel could be used to remove the overpaint successfully without damaging the original paint. Pemulen is a polyacrylic acid copolymer used in the cosmetics industry to create oil-in-water emulsions for products such as sunscreen. During testing, the percentage of benzyl alcohol in the gel was gradually increased from 1% to 5% until the desired working efficacy was achieved. Benzyl alcohol was selected as a solvent because it had a minor effect on the overpaint as a free solvent during testing. Pemulen TR2 was selected as a gelling agent because it can be used to emulsify benzyl alcohol without the addition of a surfactant (unlike other materials such as Carbopol-Ethomeen gels, which require a large amount of surfactant), which is at risk of being left behind on the surface after cleaning. The Pemulen gel is likely effective here for a combination of reasons—the benzyl alcohol component of the gel works to swell the lead-based oil paint, while the aqueous components of the gel pass through the overpaint to swell and solubilize the uneven varnish layer below, creating an undercutting effect. Removing the overpaint to uncover the original sky was a slow process performed under the microscope (Figure 13). Subtle details were revealed through cleaning including a flock of small birds in flight above the landscape at right and the feathered edges of the larger bird’s forms, features that allow them to sit more convincingly in space. The large, amorphous clouds added above the birds at the center of the sky were also removed, restoring a sense of diagonal movement through the picture. The sky was likely overpainted to mask the disturbing effects of lead soap formation visible across the surface as short, dark horizontal lines (67–70, 79). After cleaning, the picture was varnished and retouched. Varnishing saturates the paint, which makes it easier to match colors during retouching, in addition to providing an isolating layer between the original paint surface and the restoration material. Paraloid B72 was selected as an isolating varnish because it resulted in the most regular, even surface appearance during testing (80). It is a thermoplastic acrylic resin comprised of an ethyl methacrylate (70%) and methyl acrylate (30%) copolymer. In theory, the varnish creates a high-molecular weight “floor” on top of which other, more saturating varnishes can be applied to build up the desired appearance. B72 is fairly insoluble in many of the solvents commonly used for retouching, which keeps multiple options open for which retouch paints can be used later in the treatment.

43 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Figure 13. During overpaint removal in the sky; the bright blue overpaint is intact at left and has been removed at right to reveal the original gray-blue sky. (see color insert) Areas of loss and abrasion were retouched using a combination of Golden PVA conservation paints, and dry pigments with Mowilith 20 in ethanol as a medium (81, 82). Because the painting was anticipated to hang at a relatively high location in the gallery, a conscious attempt was made to not retouch every damage and disturbing feature visible at close range. There are few discrete losses to the original paint; however, in many areas, the legibility of the image was disrupted by abrasion or increasing transparency of the upper paint layers. Extensive retouching was required to reintegrate the dark lines in the damaged sky; all evidence of the lines was not erased through retouching in order to maintain the picture’s naturally aged patina. Small 1 cm x 2 cm areas of discolored varnish and blue overpaint were left intact as a reference at the edge, hidden from view by the frame. The treatment restored balance within the picture, allowing the birds to feature prominently in the foreground with the landscape receding into the background. Cleaning revealed Hondecoeter’s virtuoso rendering of the birds and his energetic brushwork in the feathers. Uncovering the original gray-blue sky greatly improved the overall appearance of the picture.

Conclusion A combination of art historical research, art conservation methodology, and scientific examination has shed new light on Hondecoeter’s Cock and Hens in a Landscape. Important, yet obscured iconographical elements were recovered, 44 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

and the tonal balance within the painting was restored to the extent possible. The picture is now on permanent display in the Prince William V Gallery, a satellite gallery of the Mauritshuis, which is hung according to 18th-century salon style, with paintings stacked from floor to ceiling. Extensive collaboration and interdisciplinary work (as revealed, in part, by the depth of the Acknowledgments section) enables powerful new insights into cultural heritage objects to be achieved.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

Acknowledgments Lauren Bradley would like to thank the American Friends of the Mauritshuis for generously funding the Mauritshuis Internship in Paintings Conservation and the Samuel H. Kress Foundation for awarding a travel grant to relocate to The Netherlands. Mauritshuis Conservators Petria Noble, Sabrina Meloni, and Carol Pottasch provided excellent mentorship and were the source of many insightful discussions throughout this project and during the internship as a whole. Conservation Scientist Annelies van Loon assisted with the interpretation of cross-section samples and fellow Mauritshuis intern Charlotte Blachon offered words of encouragement. Mauritshuis Curators Edwin Buijsen, Quentin Buvelot, Geerte Broersma, Lea van der Vinde, and Director Emilie Gordenker visited the studio on a regular basis to see the painting and provide fresh insights. Curators and conservators at other institutions graciously pulled files and brought paintings out of storage for examination including Barbara Schoonhoven at the Museum Boijmans Van Beuningen, Anna Krekeler at the Rijksmuseum Amsterdam, and Lloyd de Witt at the Philadelphia Museum of Art. Amsterdam art dealer, Pieter de Boer, kindly granted access to three Hondecoeter paintings at his gallery in addition to opening his personal photographic archives for study. A special thank you to Fred Meijer of the RKD and Joy Kearney for visiting the Mauritshuis to discuss the painting after treatment and for helping to enrich our understanding of Hondecoeter’s work. WUDPAC faculty advisor Joyce Hill Stoner offered tireless guidance and support in addition to Mary McGinn, Richard Wolbers, and Stephanie Auffret. Additional gratitude is expressed to the following institutions for providing assistance to WUDPAC students: American Institute for Conservation, Foundation of the American Institute for Conservation, The Samuel H. Kress Foundation, The National Endowment for the Humanities, The Rosenberg Family, The Society of Winterthur Fellows, Tru Vue, and The WUDPAC Professional Development Fund. The NSF is gratefully acknowledged for recently funding non-destructive analytical instrumentation (NSF 0959625) at W&L that was used in the examinations reported here; the NSF is also thanked for funding Chemistry in Art Workshops for many years. W&L Lenfest summer grants and travel grants to ESU permitted course development and the research travel that enabled the handheld pXRF to be taken to The Mauritshuis. A 2009 State Council of Higher Education for Virginia Outstanding Faculty Award assisted ESU in various aspects of these projects. ESU would like to thank Petria Noble, Carol Pottasch, and Sabrina Meloni of The Mauritshuis and Jennifer Mass and her colleagues at Winterthur 45 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

for years of collaborative help; and he would like to thank Dr. Bruce J. Kaiser for pXRF support. ESU does not have the space here to individually thank all of the W&L, U.S., and Dutch friends and colleagues who have helped support his work.

References

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

1.

2. 3. 4. 5.

6.

7.

8.

9. 10. 11. 12. 13. 14. 15. 16.

17.

The reader should note that because this is not a review paper, the references are not intended to be comprehensive. Rather, they are provided with the intention of guiding readers new to the field of paintings conservation and technical art investigations towards relevant literature. van der Vinde, L.; Bradley, L. Mauritshuis in Focus 2011, 24 (1), 32. Uffelman, E. S. J. Chem. Educ. 2007, 84, 1617–1624. Uffelman, E. S. J. Chem. Educ. (online paper) 2007, 84, 38. Uffelman, E. S. International Council of Museums Conservation Committee Triennial 16th Conference, Lisbon, Portugal, September 19−23, 2011; Criterio: Lisbon, Portugal, 2011; paper 312; pp 1−8. Hill, P., S.; Simon, D.; Uffelman, E. S.; Bower, N.; Lagalante, A. F.; Norbutus, A. J. American Institute for Conservation of Historic and Artistic Works 39th Annual Meeting, Philadelphia, PA, 2011. Code of Ethics and Guidelines for Practice of the American Institute for Conservation of Historic & Artistic Works. In American Institute for Conservation of Historic and Artistic Works: Directory 2011; AIC: Washington, DC, 2011; pp 282−288. Pinna, D.; Galeotti, M.; Mazzeo, R. Scientific Examination for the Investigation of Paintings. A Handbook for Conservator-restorers; Centro Di della Edifimi srl: Firenze, Italy, 2009. Bomford, D. Art in the Making: Underdrawings in Renaissance Paintings; National Gallery (London) Company: London, 2002. Ricciardi, P.; Delaney, J. K.; Glinsman, L.; Thoury, M.; Facini, M.; de la Rie, R. Proc. SPIE 2009, 7391, 739106-1–739106-12. Taft, W. S.; Mayer, J. W. The Science of Paintings; Springer-Verlag: New York, 2000. Trentelman, K.; Bouchard, M.; Ganio, M.; Namowicz, C.; Patterson, C. S.; Walton, M. X-Ray Spectrometry 2010, 39, 159–166. Namowicz, C.; Trentelman, K.; McGlinchey, C. Powder Diffr. 2009, 24, 124–129. Shugar, A. N.; Mass, J. L. Handheld XRF for Art and Archaeology; Lueven: Belgium, 2012. Kirsh, A.; Levenson, R. S. Seeing Through Paintings; Yale University Press: New Haven, 2000. Appendix G: Polarized Light Microscopy in Conservation. In Microscopy for Art Conservators: Appendices; McCrone Research Institute: Chicago, 2005. Goldstein, J. I.; Newbury, D. E.; Joy, D. C.; Lyman, C. E.; Echlin, P.; Lifshin, E.; Sawyer, L.; Michael, J. R. Scanning Electron Microscopy and X-ray Microanalysis; Springer: New York, 2003. 46

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

18. Hsu, C.-P. S. In Handbook of Instrumental Techniques for Analytical Chemistry; Settle, F., Ed.; Prentice Hall PTR: Upper Saddle River, NJ, 1997; pp 247−284. 19. Derrick, M. R.; Stulik, D.; Landry, J. M. Infrared Spectroscopy in Conservation Science; The Getty Conservation Trust: Los Angeles, 1999. 20. Kearney, J. Melchior d’Hondecoeter: Catalogue Raisonne 2012 in press. 21. Haak, B. The Golden Age: Dutch Painters of the Seventeenth Century; Stewart, Tabori and Chang Publishers: New York, 1996. 22. Rikken, M. Melchior d’Hondecoeter Bird Painter; Nieuw Amsterdam Publishers: Amsterdam, 2008. 23. Spicer, J. A.; Muhlberger, R. C. Hondecoeter, d’. . http://www. oxfordartonline.com/subscriber/article/grove/art/T038750 (March 26, 2012). 24. Schama, S. Embarrassment of Riches: An Interpretation of Dutch Culture in the Golden Age; Vintage Books: New York, 1997. 25. Benedict, P. Christ’s Church Purely Reformed: A Social History of Calvanism; Yale University Press: New Haven, 2002. 26. North, M. Art and Commerce in the Dutch Golden Age; Yale University Press: New Haven, 1997. 27. Hochstrasser, J. B. Still Life and Trade in the Dutch Golden Age; Yale University Press: New Haven, CT, 2006. 28. Bernstein, W. J. A Splendid Exchange: How Trade Shaped the World; Atlantic Monthly Press: New York, 2008. 29. Brook, T. Vermeer’s Hat: The Seventeenth Century and the Dawn of the Global World; Bloomsbury Press: New York, 2008. 30. Cook, H. J. Matters of Exchange: Commerce, Medicine, and Science in the Dutch Golden Age; Yale University Press: New Haven, 2007. 31. Schwartz, G. The Rembrandt Book; Harry N. Abrams, Inc.: New York, 2006. 32. Schneider, N. Still Life; Taschen: Koln, Germany, 2003. 33. Biesboer, P.; Brunner-Bulst, M.; Gregory, H. D.; Klemm, C. Pieter Claesz: Master of Haarlem Still Life; Waanders: Zwolle, The Netherlands, 2004. 34. Rosenberg, J.; Slive, S.; Kuile, E. H. t. Dutch Art and Architecture 16001800: The Pelican History of Art; Penguin Books: New York, 1979. 35. Wheelock, A. K., Jr. Aelbert Cuyp; Thames & Hudson: New York, 2001. 36. Kloek, W. T. Aelbert Cuyp: Land, Water, Light; Waanders Printing: Zwolle, The Netherlands, 2002. 37. Bruyn, J. In Rembrandt: the Master & his Workshop: Paintings; Brown, C., Kelch, J., Thiel, P. v., Eds.; Yale University Press: New Haven, 1991; pp 68−89. 38. Walsh, J. Jan Steen: The Drawing Lesson; Getty Museum: Los Angeles, 1996. 39. Dardes, K.; Rothe, A. The Structural Conservation of Panel Paintings; Getty Conservation Institute: Los Angeles, 1998. 40. van de Wetering, E.; Franken, M.; Groen, K.; Klein, P.; van der Veen, J.; de Winkel, M. A Corpus of Rembrandt Paintings V: Small Scale History Paintings; Springer: Dordrecht, The Netherlands, 2011; Vol. IV. 47 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

41. Wetering, E. v. d. Rembrandt: The Painter at Work; Amsterdam University Press: Amsterdam, 1997. 42. Doherty, T.; Woollett, A. T. Looking at Paintings: A Guide to Technical Terms Revised Edition; The J. Paul Getty Museum: Los Angeles, 2009. 43. Woll, A. R.; Mass, J.; Bisulca, C.; Cushman, M.; Griggs, C.; Wazny, T.; Ocon, N. Stud. Conserv. 2008, 53, 93–109. 44. Uffelman, E. S.; Court, E.; Marciari, J.; Miller, A.; Cox, L. Handheld XRF Analyses of Two Veronese Paintings. In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P. L., Armitage, R. A., Eds.; ACS Symposium Series 1103; American Chemical Society: Washington, DC, 2012; Chapter 3. 45. van der Doelen, G. A., Ph.D. Thesis, FOM-Institute of Atomic and Molecular Physics, Amsterdam, The Netherlands, 1999; 178 pages. 46. Doelen, G. A. v. d.; Berg, K. J. v. d.; Boon, J. J. Stud. Conserv. 1998, 43 (4), 249–264. 47. De Wild, D. Mauritshuis Conservation Treatment Report, Mauritshuis Conservation Department; Den Haag, The Netherlands, 1916. 48. Traas, J. C. Mauritshuis Conservation Treatment Report, Mauritshuis Conservation Department; Den Haag, The Netherlands, 1937. 49. de la Rie, E. R.; Delaney, J. K.; Morales, K. M.; Maines, C. A.; Sung, L.-P. Stud. Conserv. 2010, 55 (2), 134–143. 50. Delaney, J. K.; de la Rie, E. R.; Elias, M.; Sung, L.-P.; Morales, K. M. Stud. Conserv. 2008, 53 (3), 170–186. 51. Berns, R. S.; De la Rie, E. R. Stud. Conserv. 2003, 48 (4), 251–262. 52. Berns, R. S.; De la Rie, E. R. Stud. Conserv. 2003, 48 (2), 73–82. 53. de la Rie, E. R. Stud. Conserv. 1987, 32 (1), 1–13. 54. Nicolaus, K. The Restoration of Paintings; Konemann: Cologne, 1998. 55. Mauritshuis Inventory Book, Mauritshuis Curatorial Department; Den Haag, The Netherlands, 1876. 56. Wadum, J. Vermeer Illuminated: Conservation, Restoration and Research; V&K Publishing/Inmere: Mauritshuis, The Hague, 1995. 57. Buckley, B. Paintings Conservation Catalog: Stretchers and Strainers; American Institute for Conservation of Historic and Artistic Works: Washington, DC, 2008; Vol. 2. 58. Noble, P.; Meloni, S.; Pottasch, C.; Ploeg, P. v. d. Preserving our Heritage: Conservation, Restoration and Technical Research in the Mauritshuis; Waanders Publishers: Zwolle, The Netherlands, 2009. 59. Berrie, B. H. In Artists’ Pigments: A Handbook of Their History and Characteristics; Fitzhugh, E. W., Ed.; Oxford University Press: New York, 1997; Vol. 3, pp 191−272. 60. Eastaugh, N.; Walsh, V.; Chaplin, T.; Siddall, R., Pigment Compendium: A Dictionary and Optical Microscopy of Historical Pigments. ButterworthHeinemann: Oxford, 2008. 61. Kirby, J.; Saunders, D. In National Gallery Technical Bulletin; Roy, A., Ed.; National Gallery Company: London, 2004; Vol. 25, pp 77−99.

48 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by COLUMBIA UNIV on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch002

62. Muhlethaler, B.; Thissen, J. In Artists’ Pigments: A Handbook of Their History and Characteristics; Roy, A., Ed.; Oxford University Press: New York, 1993; Vol. 2, pp 113−130. 63. Hommes, M. v. E. Changing Pictures: Discoloration in 15th-17th Century Oil Paintings; Archetype: London, 2004. 64. Boon, J. J.; Keune, K.; Geldof, M.; Mensch, K.; Bryan, S.; van Asperen de Boer, J. R. J. Chimia 2001, 55, 952–960. 65. Spring, M.; Higgitt, C.; Saunders, D. In National Gallery Technical Bulletin; Roy, A., Ed.; National Gallery/Yale University Press: London, 2005; Vol. 26, pp 56−70. 66. Robinet, L.; Spring, M.; Pages-Camagna, S.; Vantelon, D.; Trcera, N. Analytical Chemistry 2011, 83, 5145–5152. 67. Noble, P.; van Loon, A.; Boon, J. J. In Chemical changes in old master paintings II: darkening due to increased transparency as a result of metal soap formation; ICOM-CC 14th Triennial Meeting, The Hague, 2005; Verger, I., Ed.; James & James: The Hague, 2005; pp 496−503. 68. Boon, J. J.; Hoogland, F.; Keune, K. In Chemical Processes in Aged Oil Paints Affecting Metal Soap Migration and Aggregation; 34th Annual Meeting of the American Institute for Conservation of Historic and Artistic Works, June 16-19, 2006, Providence, RI, 2006; pp 1−25. 69. Keune, K. Ph.D. Thesis, FOM-Institute of Atomic and Molecular Physics, Amsterdam, The Netherlands, 2005; 301 pages. 70. Robinet, L.; Corbeil, M.-C. Stud. Conserv. 2003, 48 (1), 23–40. 71. Loon, A. v. AMOLF-FOM; Amsterdam, 2008; p 234 (notes 24 and 25). 72. Wolbers, R. Cleaning Painted Surfaces: Aqueous Methods; Archetype: London, 2000. 73. Dorge, V. Solvent Gels for the Cleaning of Works of Art: The Residue Question; The Getty Conservation Institute: Los Angeles, 2004. 74. Science for Conservators Series: Cleaning, 2nd ed.; Routledge: Florence, KY, 1992; Vol. 2. 75. White, R.; Roy, A. Stud. Conserv. 1998, 43, 159–176. 76. Phenix, A.; Sutherland, K. Rev. Conserv. 2001, 2, 47–60. 77. Morrison, R.; Abigail, B.-Y.; Burnstock, A.; Van Den Berg, K. J.; Van Keulen, H. Stud. Conserv. 2007, 52 (4), 255–270. 78. Mills, J. S.; White, R. The Organic Chemistry of Museum Objects, 2nd ed.; Butterworth-Heinemann: London, 1994. 79. Keune, K.; Boon, J. J. Stud. Conserv. 2007, 52, 161–176. 80. Ellison, R.; Smithen, P.; Turnbull, R. Mixing and Matching: Approaches to Retouching Paintings; Archetype Publications Ltd.: London, 2010. 81. Berger, G. A. In Inpainting using PVA medium; Cleaning, Retouching and Coatings, preprints of the Contributions to the IIC Brussels Congress, Brussels, 1990; pp 150−155. 82. Cove, S. In Mixing and Matching: Approaches to Retouching Paintings; Ellison, R., Smithen, P., Turnbull, R., Eds.; Archetype Publications Ltd.: London, 2010; pp 74−86.

49 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Chapter 3

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Handheld XRF Analyses of Two Veronese Paintings Erich Stuart Uffelman,*,1 Elizabeth Court,*,2 John Marciari,*,3 Alexis Miller,2 and Lauren Cox2 1Department

of Chemistry, Washington and Lee University, Lexington, VA 24450 2Balboa Art Conservation Center, P.O. Box 3755, San Diego, CA 92163-1755 3The San Diego Museum of Art, P.O. Box 122107, San Diego, CA 92112 *E-mail: [email protected]; [email protected]; [email protected]

Paolo Veronese’s painting of Apollo and Daphne at the San Diego Museum of Art was suspected of containing large areas of degraded smalt (a pigment derived from ground cobalt glass) in the sky, causing the original blue color to have turned gray. Handheld XRF analysis of the painting confirmed the presence of cobalt in all spots involving the sky. For purposes of contrast, Veronese’s painting of Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine at The Timken Museum of Art was analyzed by handheld XRF. The Timken’s Veronese’s sky is still blue, and copper (almost certainly present in the form of azurite) was found instead of cobalt.

Introduction One of the themes of this book is to illustrate the cooperative efforts between scientific, academic, and museum communities in gaining new knowledge about cultural heritage material and using it to educate both the general public and students at the undergraduate and graduate levels. This chapter arose from collaborations between the Balboa Art Conservation Center and Washington and Lee University that date to 2004. Elizabeth Court was brought to W&L from BACC, with funds from an Associated Colleges of the South Keck Foundation Grant, to consult on Uffelman’s courses involving the technical examination of 17th-century Dutch painting (1–3). In the spring of 2011 an American Chemical © 2012 American Chemical Society In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Society Meeting in Anaheim, CA featuring a session on art and analytical chemistry (4) made further travel to San Diego with W&L’s handheld (portable) X-ray fluorescence spectrometer (pXRF) an efficient proposition. Two days of analyses of paintings at BACC, the San Diego Museum of Art, and the Timken Museum of Art ensued. Among the works investigated were two paintings by Veronese: Apollo and Daphne (Figure 1) at the SDMA, and the Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine (Figure 2) at the Timken. This monograph is thus intended not only as a contribution to the art conservation literature, but is also intended to be useful in various undergraduate courses on Chemistry in Art (2, 3, 5), as well as the NSF Chemistry in Art Workshops (5, 6). [Please also see chapters by Lang; Gaquere-Parker and Parker; and Hill in this volume.] Thus, we briefly discuss Veronese’s place in 16th century Italian Painting, the art historical context of the paintings, and the results of their technical examination. Since smalt degradation was a question surrounding the analysis of Apollo and Daphne, we also briefly review the degradation and detection of smalt as a colorant.

Background Paolo Veronese Paolo Caliari (1528–1588), better known as Paolo Veronese, was born in Verona, the son of a Lombard stonemason, and was apprenticed by the age of ten to the local painter Antonio Badile; he later trained with the painter Giovanni Caroto, also from Verona. Coming of age, Veronese quickly surpassed his local contemporaries, and by the early 1550s had moved to Venice, the most important artistic center in Northern Italy. In Verona, he had learned the principles of design associated with the artists of central Italy, and to these, upon arriving in Venice, he added the lessons in color available in the works of Titian. Titian, however, was for the most part engaged in work for foreign clients by the 1550s, and Veronese quickly began to win major commissions in Venice, rivaled only by his contemporary, Tintoretto. During a long and successful career, in which Veronese and his well-trained workshop turned out hundreds of paintings, the artist came to be known for his clarity of composition, bright colors, and lush painterliness. With his frescoes and large canvases, Veronese was the decorative painter par excellence. The best single introduction to Veronese in English remains the catalogue for the exhibition at the National Gallery of Art in Washington (7). See also more recent treatments (8–10). Apollo and Daphne Veronese’s canvas of Apollo and Daphne in the SDMA was painted in the early 1560s. Nothing is known of the work’s early history, but it was very likely to have been part of a decorative scheme with another mythological canvas, the Diana and Actaeon in the Philadelphia Museum of Art. Not only do the two paintings have similar subjects drawn from Ovid’s Metamorphoses, but both have figures and trees drawn to the same scale, and the two paintings are executed 52 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

in virtually identical technique: both are on a rough canvas of approximately 10 x 10 threads per square centimeter, with a thinly applied brown ground over which the forms were built up in opaque paint, with glazes (thin transparent paint layers applied over an opaque paint layer (11)) laid over the top. These and other mythological paintings by Veronese from the late 1550s and early 1560s are generally thought to be a response to the poesie that Titian painted for Philip II during the 1550s. Moreover, a group of Veronese’s mythological paintings follow close on the heels of his secular decorative ensembles at the Palazzo Trevisan in Murano (1557-58) and the Villa Barbaro at Maser (1560-61).

Figure 1. Paolo Veronese, Apollo and Daphne. Oil on canvas, H. 43 1/16” (109.4 cm) x W. 44 5/8” (113.4 cm). Gift of Anne R. and Amy Putnam to the San Diego Museum of Art, inv. no, 1945:27. Courtesy of the San Diego Museum of Art. (see color insert) 53 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

54 Figure 2. Paolo Veronese, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine. Oil on canvas, H. 40 5/8” (103.2 cm) x W. 61 ¾” (156.9 cm). Courtesy of the Putnam Foundation, Timken Museum of Art, San Diego, CA. (see color insert)

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

There are at least two textual versions of the Apollo and Daphne story (12), but the version that had the greatest impact on European art is from Ovid. In that version, Apollo saw Cupid, son of Venus, drawing his bow and insulted him, “What are you doing, you silly boy, with such a powerful weapon? A bow like that looks better on my shoulder, for I can hit anything---any animal I hunt, any enemy---in fact, just now I killed the poison-bloated Python with more arrows than I can count…. Be satisfied to arouse a little passion now and then with your torch, you certainly should not consider yourself in the same class as me” (13)! In response to this provocation, Cupid shot Apollo through the heart with a sharp, gold-tipped arrow and then shot Daphne, a river nymph, through the heart with a blunt, lead-tipped arrow. The consequence was that Daphne rejected the love of all men and sought to emulate the chaste goddess Diana, while Apollo lusted for Daphne. Apollo pursued the Nymph, who fled. Approaching exhaustion, she reached the waters of her father, Peneus. “Father, help me! If rivers have divine power---this beauty that has made me so attractive, rid me of it, change me” (13)! As her prayer concluded, she became encased in bark, her hair became leaves and her arms tree branches. Apollo, still love-struck, tried to kiss the tree, which even rooted to the ground, resisted his advances. Finally, Apollo declared that if she cannot be his love, she will be his tree, and he declared that he will always wear laurel leaves and that a laurel wreath will, from that day forth, be worn by Roman victors. The story of Apollo and Daphne is among the most familiar tales of Greco-Roman mythology and a favorite subject for artists in Renaissance and Baroque Europe. The best known version is surely Bernini’s sculpture in the Villa Borghese at Rome, but other important examples include Antonio del Pollaiuolo’s famous painting (ca. 1470-1480) in the National Gallery (London); Nicolas Poussin’s painting (1625) in the Alte Pinakothek (Munich); and Giambattista Tiepolo’s painting (ca. 1744-1745) at the Louvre. All of these works focus on the dramatic point in which Daphne is in mid-tranformation. Executed in the painterly style so characteristic of Veronese’s work, the San Diego canvas presents an interesting take on the story, for it shows not the climactic moment of the chase and Daphne’s transformation, but rather, the anticlimactic moment just afterwards: Apollo has overtaken Daphne and stares upward with wonder at the metamorphosis that has occurred. Daphne would have been running from the right towards the left of the scene and staring backward at a pursuing Apollo before her transformation; she is now frozen in that pose, and Apollo has circled around to the front of her and gazes at the ongoing metamorphosis. Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine Veronese’s painting at the Timken dates to around 1568-70, slightly later than Apollo and Daphne. The identification of the figures in the canvas has not always been clear, for the saint at right lacks Saint Catherine’s traditional attributes of a broken wheel or a ring being exchanged with the infant Christ, and she was long identified as Saint Justina. The painting seems to be based, however, on sketches that are today in the Museum Boijmans van Beuningen in Rotterdam, 55 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

and the inscription on that sheet, “Sposi di Santa […] / con S. Zuani …” (Marriage of St. ___, with St. John) puts an end to the question and confirms that the female figure at right in the painting is indeed Saint Catherine, for there would be no reason to show Saint Justina in a marriage scene. The Rotterdam drawing was made by Veronese on the back of a letter dated 1568, providing a terminus post quem for the painting. The Timken painting is not, however, the only canvas that relates to the Rotterdam drawing, for the sketches also seem preparatory to paintings at the Musée des Beaux-Arts, Brussels; the Musée Fabre, Montpelier; and the New Orleans Museum of Art. This group of paintings shows, furthermore, Veronese’s workshop in its full burst of production. On the one hand, the Timken painting was surely of Veronese’s invention and has deftly painted passages that show it was at least partly executed by the master himself. Yet, some parts of the Timken painting—the painting of St. Catherine’s draperies, for example—are handled in a slightly abbreviated manner that is often the benchmark of workshop assistance, and the existence of the other versions likewise raise the question of how much of any one painting Veronese executed himself. It can be argued that an overwhelming majority of the paintings that emerged from the workshop from around 1570 onward were created through a similarly collaborative process. [For further commentary on Veronese and his workshop, see the (forthcoming) catalogue related to the Veronese exhibition to be held at the Ringling Museum of Art in 2012.] When assisting a master painter, however, the workshop assistants would adopt the same materials and techniques as the master, so the Timken painting, like the canvas at the San Diego Museum of Art, can be considered a typical work by Veronese from the 1560s, particularly from a technical point of view.

Degradation and Detection of Smalt Colorants The exact date of cobalt as a colorant in glass is lost in antiquity, and the use of cobalt glass, smalt, as a pigment in painting begins prior to the 14th century (14). Its production involved melting silica with potash (potassium oxide) and a source of cobalt derived from ores that were frequently rich in arsenic, as well as containing other metals such as iron and nickel. The amount of other elements present depended on both the source of the ore and its treatment; i.e., roasting the ore lowered the arsenic content via volatilization. Smalt’s use in European painting was particularly prevalent in the 16th and 17th centuries, because it was a much cheaper alternative to azurite and ultramarine. With the advent of Prussian blue in the early 18th century, smalt’s use became increasingly infrequent. Although smalt yields a pleasing blue color when fresh, because the refractive index of glass is not dramatically different from polymerized oil binding medium, it had to be ground fairly coarsely for reasonable opacity. In addition, the ground glass-oil medium combination did not yield ideal handling properties (e.g., the oil could run down a vertical canvas). Furthermore, as early as the 17th century, smalt was known to discolor, although the early writers attributed the discoloration to the oil binding medium, rather than the pigment itself. 56 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

In the last decade, numerous studies have elucidated the nature of smalt degradation (15–20). A combination of microspectroscopy, scanning electron microscopy energy dispersive spectroscopy (SEM-EDS), and secondary ion mass spectrometry (SIMS) demonstrated that loss of potassium ions from the glass was linked to a change in smalt from blue to gray (15). Cobalt ions in blue glass are found in a tetrahedral +II oxidation state, and the cobalt in this environment produces an absorption spectrum entirely consistent with a d7 Td Co(II) center in a weak ligand field. The basicity of the glass is essential to stabilizing the Td coordination of Co(II) within the glass framework, and the potassium acts as a counterion to balance the charge; use of sodium to balance the charge (soda glass) produces a spectrum that has a bathochromic shift that gives the material a less-pleasing purplish appearance. As potassium leaches out of the glass and is replaced by protons, the coordination number of the Co(II) increases to Oh six-coordinate, and consistent with this change in the coordination environment is the shift in the absorption spectrum and the dramatic change in aborbance power due to the alteration from local point group symmetry without an inversion center (Td) to a local point group symmetry with an inversion center (Oh). SEM-EDS maps and SIMS maps consistently showed that cobalt was not leaching from the glass and that potassium was, yielding potassium soaps from ester saponification of the oil binding medium (15). The alteration of the glass caused by potassium ion leeching out into the surrounding medium can be distinguished by vibrational methods (17, 19, 21). In smalt, the asymmetric Si-O-Si stretch occurs in the region 1040-1080 cm-1; the symmetric Si-O-Si stretch occurs in the region of 780-800 cm-1; and the OSi-O bending vibration occurs between 460-470 cm-1. Crucially, additives such as Na2O, CaO, and K2O generate an additional vibration at approximately 920 cm-1, corresponding to Si-OM (where M = Na+, Ca2+, or K+; for potassium, this additional band is generally underneath the asymmetric Si-O-Si stretch). When the potassium leaches out of smalt, the Si-OK band is observed to diminish, and Si-OH bands are observed to grow (17, 21). The most definitive characterization of the chemical changes at cobalt in smalt degradation has recently been published (20). By using the SOLEIL synchrotron, investigators were able to examine degraded and undegraded smalt samples at very high resolution using micro-X-ray absorption spectroscopy (μ-XAS) at the cobalt K-edge. By employing X-ray absorption near-edge spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS), as well as synchrotron-based μ-FTIR and μ-Raman, they were able to probe the different alterations. By using anhydrous blue Co(II)-silica gel and hydrated pink Co(II)-silica gel as model compounds in XANES and EXAFS measurements of degraded and undegraded smalt, they were able to conclusively show that the well-preserved smalt featured Co(II) in a Td environment with Co-O bond distances of 1.95 Å, while the degraded smalt featured Co(II) in an Oh environment with Co-O bond distances of 2.02-2.06 Å. Although their results are generally applicable, it is noteworthy in this context that two of the five paintings studied by these synchrotron methods featured degraded and well-preserved smalt samples from works by Veronese. 57 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Fortunately, in addition to the investigations of smalt referenced above, the National Gallery, London, published two significant papers on Veronese’s technique, spanning most of his career, based on his nine paintings in their collection (22, 23), and these papers document Veronese’s use of smalt in the sky of some works and the use of azurite in the sky of other works. The technical results were recently placed in a larger context of Veronese’s work (24). Veronese’s reputation as a master of color was facilitated by his location in Venice, which was a trade nexus and would have given him access to all of the pigments available at that time in Europe. Venetian glass was renowned, and would have meant that smalt was readily available to Veronese. The National Gallery work demonstrated that, in several paintings featuring gray-toned skies, the color was not the result of an artistic choice by Veronese, but rather the degradation of smalt over time. Furthermore, Veronese’s green copper resinates exhibit highly variable degrees of preservation/degradation, and his significant use of red lake pigments (characteristic of Venetian painting of the period) has caused many of his reds to be fugitive. A very readable reference on discoloration in paintings of this era is available (25). Use of pXRF for the Analysis of Cultural Heritage Objects Handheld XRF has several pros and a few cons as an analytical tool for examining cultural heritage objects (26, 27). Some of the pros are: (1) Because the instrument is portable, the instrument may be brought to the object, eliminating the chance of damage due to transport. [Of course, there is a physical risk of damage, that must be controlled, in bringing the instrument into close proximity to the object.] For paintings, as long as the artwork is not covered by glass, the instrument may be mounted on a tripod and brought directly to the painting in the gallery. (2) Significantly, the technique is non-destructive; i.e., no sample needs to be removed from the art object. (3) The amount of X-ray radiation emitted is so low that the art object will be undamaged by the analysis. Furthermore, the emitted radiation is so low that users who exercise prudent caution should never be exposed to radiation above background levels. (4) Because X-rays have penetrating power, layers below the surface of the object yield information to the analyst. (5) With the use of a portable vacuum pump (to evacuate the interior of the instrument’s X-ray path) and with careful positioning of the instrument within one millimeter of the object, elements as light as aluminum can be detected (air absorbs X-rays of low energy fluoresced by the lighter elements). (6) Analyses can be performed rapidly. Two minutes of acquisition time is sufficient in many instances to get good quality signal to noise ratios for most peaks of interest. When dealing with valuable cultural heritage objects, it frequently takes longer to carefully position the spectrometer than it does to acquire the data. (7) The instrument is relatively inexpensive and is physically robust to travel in the field. There are cons or limitations associated with the technique. (1) Unlike benchtop pXRF models which can have sampling resolutions of 70 μm, the handheld unit has a sampling area of roughly 3 mm in diameter. This can necessitate caution in selecting areas to analyze for clear results. (2) Although 58 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

it is useful to get information from different layers of the painting due to the penetrating power of the X-rays, determining which layer of the painting is causing which signals frequently requires having other forms of information; e.g., paint samples or cross sections analyzed by polarized light microscopy (PLM) or SEM-EDS. Recently, 3D resolution and mapping of pXRF spectra have been achieved by synchrotron methods (28–31), or very advanced semi-portable systems (32). (3) Quantitative information is difficult to obtain and is typically impossible to obtain for paintings. (4) The instrument’s construction gives rise to element signals that sometimes cause issues with correct analysis. For instance, the instrument construction typically produces significant Rh peaks (for a rhodium anode X-ray source) as well as low level peaks for Fe, Co, and Ni (from stainless steel), and Al (instrument structure) (33). With experience, and sufficient authentic material in the sample area, real peaks from these elements in the sample can be distinguished from the intensity of the peaks present from those elements in the instrument. There are also sum peaks, escape peaks, elastic scattering of X-rays, inelastic scattering of X-rays, and other artifacts to consider (26, 33). Interested readers are referred to a significant new book in press on the use of handheld pXRF for art and archaeology applications (34), and a recent well-illustrated general survey of scientific methods for examining paintings (35).

Experimental Section For the pXRF spectra obtained in this study, the following conditions were typically used: Bruker Tracer III-SD, Rh X-ray tube at 40 kV and 11 μamps, 20 Torr interior pressure inside the instrument, 120 s spectrum accumulation time. The front of the instrument was carefully placed parallel to the picture plane. Because the painting surface was relatively flat in the regions surveyed, the instrument could be safely very carefully positioned (using a tripod with a manually controlled, geared instrument mount) within approximately one millimeter of the painting surface. The element assignments were made using the Bruker Artax 7.2.1.1 software, but comprehensive tables of pXRF lines are readily available in common sources (36).

Results of Technical Examination Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine Prior to the 2011 pXRF investigation, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine from the Timken Museum of Art was examined as part of a project at BACC in 2000-2001 to determine the materials used by the artist. The painting was examined in normal and ultraviolet light and under a stereobinocular microscope. In addition, cross sections and pigment samples were taken. The cross sections taken from the painting were useful in determining the layered structure of the painting and how Veronese applied the paint. [For an example of a paint cross section showing layering structure, see Figure 6 of the previous chapter of this volume (37).] This technique 59 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

requires extremely small samples (generally ranging 0.5-1 mm in diameter) be taken. Samples containing as complete a representation of layering as possible are removed from areas adjacent to pre-existing structural damage (38). The samples are mounted in resin, ground and polished, and examined under a microscope. PLM is used to identify the individual pigment particles that make up paint layers (18). This technique requires that extremely small samples be taken from the surface of the painting. A small scraping of the top layer of paint is mounted on a microscope slide and viewed under a microscope with light polarizing capabilities. The pigment particles are examined under high magnification to determine standard particle characteristics, such as color, morphology, transparency, birefringence/isotropism, index of refraction, particle relief, pleochroism, polarization colors, and particle size (18). It is possible to identify the pigments in the sample by knowing the particular characteristics and then by comparing them with the characteristics of reference pigment samples. This technique should be confirmed with elemental analysis whenever possible in order to make a conclusive identification, as was done with these samples using pXRF analysis. Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine is oil on canvas and measures H. 40 5/8” (103.2 cm) x W. 61 ¾” (156.9 cm). The original canvas is a coarse, herring-bone-weave linen. The herring bone pattern runs horizontally and its texture is visible on the painted surface. The original canvas is lined to a finer, plain-weave, linen fabric. Canvas lining is done to paintings when the original canvas is structurally unsound and/or there is significant loss of adhesion between the paint layers and the original canvas (39). The appearance of the paint is noticeably more colorful than the Apollo and Daphne, especially the sky which is bright blue. The canvas was prepared with a white ground layer that is probably gesso (gesso is calcium sulphate, gypsum, mixed with hide glue---typically rabit-skin glue (11)). An imprimatura or toning layer was next applied over the entire gesso ground and provides a mid-tone for the composition. The paint is built up in thin layers, and the final effect depends upon interlayering of both transparent glazes and opaque scumbles (translucent light colored paint). The flesh and the red colors were underpainted in grisaille (gray) using a mixture of white and black paint, whereas the other parts of the painting were underpainted with white, blacks, and browns. The bright colors were built up with more opaque pigment and the darker colors were built up with transparent glazes. It is the combination of the layering that achieves the beautiful surface. The deep red glazes over the shadows in the dress and the brown glazes in the dark areas give the painting depth and contrast with the thicker more opaquely painted areas such as the sky, flesh, and garment details that have been highlighted with opaque touches of white and lead tin yellow. Polarized light microscopy on several samples from the painting indicated the presence of azurite blue (rather than smalt), lead tin yellow, vermilion, and possibly red lead (18). pXRF analysis (see Figure 3 for spot locations) revealed copper peaks and confirmed the presence of azurite (Figure 4), which is a copper-containing blue pigment. The occurance of tin peaks (Figure 5) in the pXRF analysis was consistent with the presence of lead tin yellow (lead is present in all of the pXRF spectra, due to the use of lead white in the ground layer of paint). The presence of mercury and arsenic (Figure 60 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

6), was indicative of vermilion and either orpiment or realgar, respectively. Note that it was fortunate that the relative amounts of Hg and As relative to the Pb signals made the assignments possible (especially with curve fitting) even with the peak overlap issues present with these three elements (26, 36). [The Hg Lα2 and Lα1 lines at 9.90 and 9.99 keV respectively do not overlap with Pb or As lines or any other relevant element lines. The Pb Lα2 and Lα1 lines at 10.45 and 10.55 keV respectively overlap with the As Kα2 and Kα1 lines at 10.51 and 10.54 keV respectively; however, several other Pb lines in the spectrum do not overlap with any lines from elements that could reasonably be expected. The As Kβ2 and Kβ1 lines at 11.72 and 11.73 keV respectively overlap with the Hg Lβ2 and Lβ1 lines at 11.92 and 11.82keV respectively; however, perak deconvolution clearly resolved the As and the Hg.] The Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine was restored prior to its acquisition by the Timken. There are no large losses of ground or paint apparent. However, the painting has been cleaned in the past and has sustained solvent abrasion throughout the composition. The abrasion was carefully inpainted by a past restorer. The condition of the painting was assessed by the Balboa Art Conservation Center in 1996 and found to be satisfactory.

Apollo and Daphne Condition Assessment The painting of Apollo and Daphne was executed with oil paint on linen canvas. The original canvas measures H. 42 15/16” (109.1 cm) x W. 44 3/8” (112.7 cm); the overall size mounted onto the stretcher measures H. 43 1/16” (109.4 cm) x W. 44 ½” (113.4 cm). The original canvas consists of two pieces of moderately heavy (10 warp/weft threads per cm), plain-, open-weave linen fabric with threads varying in thickness, butted together and sewn with thread to make a horizontal seam at H. 31 ½”. It was common practice to sew pieces of canvas together to make a larger painting surface because fabric looms only made certain widths of canvas. There are no tacking edges of the usual type remaining on the painting, but there is some evidence that this painting, like other known examples by the artist, may have been nailed to the front face of the stretcher. Apart from a slightly irregular pattern of cusping, there seem to be now-filled holes at regular intervals along the bottom and possibly right edge(s). The canvas shows cusping at the top, right, and bottom edges, but the left seems to be trimmed to a greater extent. X-radiographs taken by the BACC helped provide insight into this issue. As noted by Penny (40), Veronese at least occasionally nailed his canvases to the front face of the stretcher. Cusping (also sometimes termed scalloping) is the pattern produced in the weave of the canvas when it is prepared for painting. The canvas threads distort into a curved pattern along the edges because they are restrained by the tacks used to stretch the canvas (38, 41, 42).

61 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

62 Figure 3. Spot locations for the pXRF analysis of Veronese, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine.

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Figure 4. Veronese, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine. Blue sky featuring prominent Cu peak (confirmed as azurite by PLM) [spot 1].

Figure 5. Veronese, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine. Bluish white and yellow dress area showing Sn (characteristic of lead tin yellow) [spot 5]. 63 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Figure 6. Veronese, Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine. Orange cloak featuring prominent Hg and As peaks (characteristic of vermilion and orpiment or realgar) [spot 2]. Before painting, the canvas was prepared with a thin, aqueous (est.) ground or priming layer that does not conceal the texture of the canvas. The ground appears to be light-brown in color, although its appearance may also be darkened by penetration of glue or discolored varnish layers during its history. The paint is generally thin and ranged in application from thin transparent glazes to moderately thick soft paste with low impasto (more thickly applied paint that may retain brush texture or a three-dimensional appearance) (11). The x-radiograph revealed that the position of Daphne’s outstretched arm and shoulder had been slightly changed as the artist worked out the composition. The sky appears to be an unusual beige or tan color rather than the expected blue. The color of the sky was one of the reasons the Curator and Conservators wanted further examination of the pigments to see whether this color was intentional or was a result of the aging of the materials. The Apollo and Daphne painting has undergone multiple campaigns of restoration over its lifetime. This is typical for older paintings as the materials age and the paintings sustain damages. Most paintings in a museum have been restored at least once or twice. The most recent conservation/restoration treatment of this painting was conducted at the Balboa Art Conservation Center from 1981-1983. At that time the painting was examined and found to have suffered from extensive damage from past harsh cleaning (solvent abrasion) that had exposed the tops of many of the canvas threads throughout the composition, but especially in the background and the sky. In the past, powerful solvents and chemical reagents with the capacity to erode the original paint were often used 64 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

to remove layers of grime, discolored varnish, and darkened paint added by previous restorers. The solvent abrasion and other damages had been covered with restorer’s paint that extended well beyond the areas of actual damage and had darkened with age. The non-original varnish was of a natural resin type that had also become yellow, and there were brownish remnants of an earlier varnish in the interstices of the weave that further detracted from the appearance of the painting. In addition to the compromised aesthetic appearance of the painting, it was structurally vulnerable. The original canvas was quite weak and had been torn in the past. It had been reinforced with two layers of linen attached to the reverse with animal skin glue perhaps 150 years earlier, but this lining had in turn become brittle and no longer gave the canvas adequate support. The treatment at BACC in 1983 consisted of both aesthetic and structural steps. The discolored varnish and restorer’s overpaint were removed to reveal the original colors. The cleaning involved using small cotton swabs to apply tailored mixtures of organic solvents that had been determined by testing under a binocular microscope to safely remove the non-original layers without further damaging the original paint. The old lining and lining adhesive were also carefully removed from the reverse, and the painting was relined to a stable synthetic fabric using a non-penetrating and reversible synthetic adhesive (BEVA 371, a proprietary poly(ethylene-co-vinyl acetate) from Adams Chemical). It was remounted onto a new self-adjusting spring stretcher. The painting was then brushed with an isolating layer of a stable synthetic varnish (Laropal K-80, a polycyclohexanone) that does not darken or turn yellow with age. [This varnish is no longer used at BACC because it may eventually cross-link. However, this is very slow under museum conditions, and it will remain safely removable in the future.] Losses were filled to the level of the original paint with gesso putty and inpainted with paint made from dry pigments mixed in the same resin to re-integrate the surface. The inpainting was confined to the actual areas of loss. Finally, a thin protective layer of the same varnish with the addition of a small amount of microcrystalline wax was sprayed on the surface to adjust the gloss. The condition of the painting is currently stable and satisfactory. [Readers new to paintings conservation might find six references particularly helpful in the context of this discussion (39, 43–47). For specific discussion of current conservation techniques and maeterials, visit the American Institute for Conservation of Historic and Artistic Works (AIC) (48).]

pXRF Analysis Test spots (see Figure 7 for spot locations) in the sky of Apollo and Daphne (Spots 1, 2, and 12) were chosen from both lighter tan and darker brown areas shown by examination under ultraviolet light to be least likely to be contaminated by later inpainting (Because older varnish and older paint fluoresce differently under UV than newer varnish and newer paint, UV-induced visible fluorescence can be used to distinguish areas of inpainting.). All these sample locations showed a similar composition (e.g., Figure 8) with cobalt, iron, and arsenic as well as silicon (from the glass), indicative of smalt. They also each contain lead, indicative of lead white mixed with the (formerly) blue paint and calcium, presumably from 65 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

the calcium sulfate ground. In this instance, the discoloration of the smalt is extreme to the point that there is no longer any blue color visible. In the case of Apollo and Daphne, harsh treatment in the past has undoubtedly contributed further. Another test (Spot 8) was done in an area of the sky near the foliage on the left where a light blue color with a greenish tinge appeared to be original paint. This area contained lead, calcium, iron, silicon, cobalt, and arsenic, but also contained copper. The ratio of the peaks in Spot 8 to the peaks of Spots 1, 2, and 12 suggest that there is less cobalt and arsenic (therefore less smalt) and more of a copper containing pigment, possibly azurite.

Figure 7. Spot locations for the pXRF analysis of Apollo and Daphne.

Other areas of the painting that would not be expected to have cobalt were chosen for analysis to provide comparison with the cobalt-containing spectra and for general reference. Spot 3 was taken from an area in the pink dress. It contained 66 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

no cobalt, but a lot of lead from lead white. The red is likely to be a lake pigment (an organic pigment not detectable with pXRF). [Sometimes the aluminum of the substrate on which the organic dye is precipitated to produce the pigment can be detected under the conditions used in this study to obtain the pXRF spectra. However detecting aluminum can be challenging because it is a light element, and it is present as part of the instrument construction. For Spot 3, we could not be confident we were detecting aluminum above background levels.] Spot 9 from an area on Apollo’s proper right thigh again contained no cobalt but lots of lead (lead white) with some iron, probably from an earth pigment. There is also a small amount of titanium (Ti), which is presumed to come from titanium white (the white pigment used for inpainting at BACC). This entire area has numerous tiny dots of inpainting that were difficult to avoid. Another test area in the sky (Spot 4, Figure 9) was chosen because it is a visual color match to the original areas found to contain smalt but is clearly an area of loss that is inpainted. This spot contained no cobalt but did contain titanium, presumed to come from titanium white, and iron, most likely from an earth pigment (ochre or umber), as well as a lot of calcium, presumably from the calcium carbonate fill. There was also some lead, although a much smaller amount than in the tests done on areas of original paint. It may be from original paint just outside the target area of inpainting or from residual original paint in the area of inpainted loss. The absence of cobalt in these samples reaffirmed its importance in the sky samples.

Figure 8. Veronese, Apollo and Daphne. Gray sky showing Co and As (both characteristic of smalt---cobalt glass) [spot 2]. 67 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Figure 9. Veronese, Apollo and Daphne. Gray sky showing Ti (anachronistic titanium white---characteristic of inpainting) [spot 4].

Figure 10. Veronese, Apollo and Daphne. Bown ivy on tree showing Cu (characteristic of discolored copper resinates) [spot 10]. 68 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

The composition of the varying greens in the painting was also of interest. There are transparent brown leaves and foliage as well as more opaque green leaves and foliage. The transparent brown foliage (Spot 10, Figure 10) and the dark brown shadow of the hill (Spot 15) have similar spectra and associated peaks for Ca, Mn, Fe, Cu, Pb. The presence of copper in these samples suggests that they may be discolored copper resinate, although the organic resinate cannot be positively identified with pXRF. The opaque light green in the leaves of the central foliage (Spot 14) was found to contain Ca, Si, Fe, Co, Cu, Pb, and As. The copper in this instance could be from a green pigment like verdigris or malachite or even the blue pigment azurite mixed with a yellow pigment such as yellow ochre, possibly indicated by the fact that there appears to be more iron than is present in the original sky samples. The presence of cobalt and arsenic also points to the use of smalt, either as an admixture or a lower layer. The medium green of the hill (Spot 16) in the background has a similar composition to the light green leaf, but the medium green seems to have a higher proportion of cobalt and a lower proportion of copper. It is possible that this area also has either verdigris, copper resinate, malachite, or azurite and yellow ochre in mixtures with smalt or layered with smalt. The brown hill (Spot 5 and Spot 13) showed a similar composition to the other browns and greens - Ca, S, Fe, Cu, Co, As, Pb – with a larger amount of iron and some detectable sulfur. This indicates the use of earth pigments, a copper pigment (verdigris, copper resinate, azurite, malachite), smalt, and lead white. The calcium and sulfur peaks suggest the presence of calcium sulfate, which would be in the ground layer. In all of the brown and green samples, since lake pigments (organic dye stuffs) cannot be detected with pXRF spectrometry, it cannot be ruled out that a yellow lake is present, possibly mixed with smalt or azurite to create green. In all these cases, other techniques such as cross-sections and polarized light microscopy could give more conclusive answers about which pigments, layered or mixed, were actually used. Lead was observed in all spots, consistent with lead white being used as an admixture in the paint layers or with lead white existing in layers beneath those analyzed. Spot 9 produced a good spectrum of lead white. Calcium was observed in all spots, consistent with calcium sulfate being used in the canvas preparation. Veronese was known to use a traditional gesso priming. The presence of manganese and iron are consistent with an earth pigment. Veronese’s technique (1570’s) often included a white ground layer with a thin brown imprimatura, traditionally made of black and earth pigments (23).

Conclusion Collaboration between BACC, SDMA, The Timken, and W&L, has led to a fruitful examination of two Veronese paintings by handheld pXRF. The work reported here confirmed a curatorial and art conservation insight at BACC and SDMA that the Veronese Apollo and Daphne has undergone extensive smalt degradation (as well as changes in other pigments). This work extends relevant work reported from the National Gallery, London, in which some of Veronese’s skies are pigmented by smalt (with varying degrees of degradation), while 69 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

others are pigmented by azurite. Combining pXRF spectroscopy with previous conservation studies by BACC of the two Veronese paintings enabled a deepened understanding of the two works. Knowing that the current appearance of the sky in Apollo and Daphne was not an artistic choice by Veronese (or a state of unfinished work), permits us to view the painting with renewed imagination, and to align Veronese’s reputation for bright colors with the current state of the painting.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

Acknowledgments The authors wish to thank John Wilson, PhD., Executive Director of the Timken Museum of Art, for graciously permitting the project team to examine the Museum’s Madonna and Child with St. Elizabeth, the Infant St. John, and St. Catherine 1565-70 for comparative purposes for this study. The NSF is gratefully acknowledged for recently funding non-destructive analytical instrumentation (NSF 0959625) at W&L that was used in the examinations reported here; the NSF is also thanked for funding Chemistry in Art Workshops for many years. W&L Lenfest summer grants and travel grants permitted course development and research travel. ESU’s travel to BACC and the SDMA was made possible by funds from the W&L Dean of the College. A 2009 State Council of Higher Education for Virginia Outstanding Faculty Award assisted ESU in various aspects of these projects. ESU would like to thank all of the staff at BACC, SDMA, and the Timken who made his research visit possible, as well as Dr. Bruce J. Kaiser for pXRF support.

References 1.

2. 3. 4.

5.

6.

7. 8.

The reader should note that, because this is not a review paper, the references are not intended to be comprehensive, but should enable the reader new to paintings investigations to get into relevant literature. Uffelman, E. S. J. Chem. Educ. 2007, 84, 1617–1624. Uffelman, E. S. J. Chem. Educ. online paper 2007, 84, 38. In Partnerships and New Analytical Methodologies at the Interface of Chemistry and Art, 241st American Chemical Society National Meeting, Anaheim, CA, March 27−31, 2011. Uffelman, E. S. International Council of Museums Conservation Committee Triennial 16th Conference, Lisbon, Portugal, September 19−23, 2011; Criterio: Lisbon, Portugal, 2011; paper 312; 1-8. Hill, P. S.; Simon, D.; Uffelman, E. S.; Bower, N.; Lagalante, A. F.; Norbutus, A. J. American Institute for Conservation of Historic and Artistic Works 39th Annual Meeting, Philadelphia, PA, 2011. Rearick, W. R. Paolo Vernese: His Life and Art; National Gallery of Art, Washington and Cambridge University Press: Cambridge, England, 1988. Cocke, R. Paolo Veronese: Piety and Display in an Age of Religious Reform; Ashgate Pub Ltd: Farnham, Surrey, U.K., 2002. 70

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

9. 10. 11. 12. 13.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

14.

15. 16. 17.

18.

19.

20. 21.

22. 23. 24.

25. 26. 27. 28.

Ilchman, F. Titian, Tintoretto, Veronese: Rivals in Renaissance Venice; MFA Publications: Boston, 2009. Salomon, X. F. Paolo Veronese: The Petrobelli Altarpiece; Silvana: Milan, 2009. Doherty, T.; Woollett, A. T. Looking at Paintings: A Guide to Technical Terms Revised Edition; The J. Paul Getty Museum: Los Angeles, 2009. Gantz, T. Early Greek Myth: A Guide to Literary and Artistic Sources; The Johns Hopkins University Press: Baltimore, MD, 1993. Simpson, M. The Metamorphoses of Ovid; University of Massachusetts Press: Amherst, MA, 2001. Muhlethaler, B.; Thissen, J. In Artists’ Pigments: A Handbook of Their History and Characteristics; Roy, A., Ed.; Oxford University Press: New York, 1993; Vol. 2, pp 113−130. Boon, J. J.; Keune, K.; Geldof, M.; Mensch, K.; Bryan, S.; van Asperen de Boer, J. R. J. Chimia 2001, 55, 952–960. van der Weerd, J.; van Veen, M. K.; Heeren, R. M. A.; Boon, J. J. Anal. Chem. 2003, 75, 716–722. Spring, M.; Higgitt, C.; Saunders, D. In National Gallery Technical Bulletin; Roy, A., Ed.; National Gallery/Yale University Press: London, 2005; Vol. 26, pp 56−70. Eastaugh, N.; Walsh, V.; Chaplin, T.; Siddall, R. Pigment Compendium: A Dictionary and Optical Microscopy of Historical Pigments; ButterworthHeinemann: Oxford, 2008. Robinet, L.; Neff, D.; Bouquillon, A.; Pages-Camagna, S.; Verney-Carron, A.; Etcheverry, M.-P.; Tate, J. In ICOM-CC 15th Triennial Conference New Delhi 22-26 September 2008 Preprints; Bridgland, J., Ed.; Allied Publishers Pvt, Ltd: New Delhi, 2008; Vol. 1, pp 224−231. Robinet, L.; Spring, M.; Pages-Camagna, S.; Vantelon, D.; Trcera, N. Anal. Chem. 2011, 83, 5145–5152. Spring, M. Investigation of the degradation of smalt in paintings using multiple synchrotron and laboratory techniques. In SR2A: Fourth edition of Synchrotron Radiation in Art and Archaeology; Amsterdam, 2010. Penny, N.; Spring, M. In National Gallery Technical Bulletin; Roy, A., Ed.; National Gallery Publications: London, 1995; Vol. 16, pp 5−29. Penny, N.; Spring, M. In National Gallery Technical Bulletin; Roy, A., Ed.; National Gallery Publications: London, 1996; Vol. 17, pp 32−55. Penny, N. National Gallery Catalogues The Sixteenth Century Italian Paintings Volume II Venice 1540-1600; National Gallery Company Limited: London, 2008; Vol. 2. Hommes, M. v. E. Changing Pictures: Discoloration in 15th-17th Century Oil Paintings; Archetype: London, 2004. Namowicz, C.; Trentelman, K.; McGlinchey, C. Powder Diffraction 2009, 24, 124–129. Trentelman, K.; Bouchard, M.; Ganio, M.; Namowicz, C.; Patterson, C. S.; Walton, M. X-Ray Spectrometry 2010, 39, 159–166. Woll, A. R.; Bilderback, D. H.; Gruner, S.; Gao, N.; Huang, R.; Bisulca, C.; Mass, J. L. In Materials Research Society Symposium Proceedings; Vandiver, 71

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

29. 30. 31. 32.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

33. 34. 35.

36.

37.

38. 39. 40.

41. 42.

43. 44. 45. 46.

P. B., Mass, J. L., Murray, A., Eds.; Materials Research Society: Warrendale, PA, 2005; Vol. 852, pp 281−290. Woll, A. R.; Mass, J.; Bisulca, C.; Huang, R.; Bilderback, D. H.; Gruner, S.; Gao, N. Appl. Phys. A 2006, 83, 235–238. Mass, J.; Bisulca, C. Antiques and Fine Art 2010Summer/Autumn, 222–223. Janssens, K.; Dik, J.; Cotte, M.; Susini, J. Acc. Chem. Res. 2010, 43 (6), 814–825. Alfeld, M.; Janssens, K.; Dik, J.; de Nolf, W.; van der Snickt, G. J. Anal. At. Spectrom. 2011, 26, 899–909. Kaiser, B. J.; Wright, A. Draft Bruker XRF Spectroscopy User Guide: Spectral Interpretation and Sources of Interference; Bruker: Coventry, 2008. Shugar, A. N.; Mass, J. L. Handheld XRF for Art and Archaeology; Lueven: Belgium, 2012. Pinna, D.; Galeotti, M.; Mazzeo, R. Scientific Examination for the Investigation of Paintings. A Handbook for Conservator-restorers; Centro Di della Edifimi srl: Firenze, Italy, 2009. Bearden, J. A. In CRC Handbook of Chemistry and Physics: A Ready-Reference Book of Chemical and Physical Data, 60th ed.; CRC Press, Inc.: Boca Raton, Florida, 1979; pp E152−E190. Bradley, L. P.; Meloni, S.; Uffelman, E. S.; Mass, J. L. Technical Examination and Treatment of a Painting by Gijsbert Gillisz d’Hondecoeter. In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P. L., Armitage, R. A., Eds.; ACS Symposium Series 1103; American Chemical Society: Washington, DC, 2012; Chapter 16. Kirsh, A.; Levenson, R. S. Seeing Through Paintings; Yale University Press: New Haven, 2000. Nicolaus, K. The Restoration of Paintings; Konemann: Cologne, 1998. Penny, N. In National Gallery Catalogues The Sixteenth Century Italian Paintings Volume II Venice 1540-1600; National Gallery Company Limited: London, 2008; Vol. 2, p 430. Wetering, E. v. d., Rembrandt: The Painter at Work; Amsterdam University Press: Amsterdam, 1997. Johnson, C. R. J.; Hendriks, E.; Noble, P.; Franken, M. In Advances in Computer-Assisted Canvas Examination: Thread Counting Algorithms; Buckley, B., Ed.; AIC Paintings Specialty Group Postprints 22: Los Angeles, 2009. Berger, G. A.; Russell, W. H. Conservation of Paintings: Research and Innovations; Archetype: London, 2000. Wolbers, R. Cleaning Painted Surfaces: Aqueous Methods; Archetype: London, 2000. Dorge, V. Solvent Gels for the Cleaning of Works of Art: The Residue Question; The Getty Conservation Institute: Los Angeles, 2004. Townsend, J. H.; Doherty, T.; Heydenreich, G.; Ridge, J. Preparation for Painting: The Artist’s Choice and Its Consequences; Archetype Publications, Ltd.: London, 2008. 72

In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 16, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch003

47. Ellison, R.; Smithen, P.; Turnbull, R. Mixing and Matching: Approaches to Retouching Paintings; Archetype Publications Ltd.: London, 2010. 48. AIC Collaborative Knowledge Base; www.conservation-wiki.com (January 10, 2012).

73 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Chapter 4

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Characterization of the Binders and Pigments in the Rock Paintings of Cueva la Conga, Nicaragua R. Li,1 S. Baker,2 C. Selvius DeRoo,3 and R. A. Armitage*,1 1Chemistry

Department, Eastern Michigan University, 501 W Mark Jefferson, Ypsilanti, MI 48197 2Archaeological/Historical Consultants, 609 Aileen St., Oakland, CA 94609 3Conservation Department, Detroit Institute of Arts, 5200 Woodward Ave., Detroit, MI 48202 *E-mail: [email protected]

Cueva la Conga is the first limestone cave with paintings and modified speleothems found in Nicaragua. Dating of images made with inorganic pigments generally requires the presence of an organic binder. Chemical characterization of the organic material in the paint was undertaken using thermally assisted hydrolysis/methylation-gas chromatography-mass spectrometry (THM-GC-MS). Results show that significant quantities of organic material are present in the rock itself, precluding dating of the paints based on binders. Some of the inorganic paints, however, contain traces of charcoal, possibly from calcination of the iron oxide pigments to change their color. We have successfully dated charcoal from the paintings using plasma-chemical oxidation and accelerator mass spectrometry. This study considered the composition of the substrate when sampling of the rock art to be dated, and emphasizes the importance of rigorous sampling protocols in analysis of rock art.

© 2012 American Chemical Society In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Introduction The Cueva la Conga rock art site in Nicaragua has only been known outside its local area since 2006, when it was recorded by archaeologists from the United States (1). The site is located in northern Nicaragua, in a region of that country where few archaeological studies have been carried out. This is due, in part, to the isolation of the area, in the mountains near the Honduran border. In 2009, we undertook a small expedition to Cueva la Conga to collect samples from the paintings found there, in hopes of determining the age of the paintings, thereby placing them into a chronological context. The cultural implications of dating the rock art of Cueva la Conga and its importance as a ritual cave are described elsewhere (2). Here, we focus on the scientific study of the materials used to create the paintings. In particular, we were interested in determining the nature of any organic material in the paints, which would aid in the radiocarbon dating of the paintings. Direct radiocarbon dating of rock paintings using plasma-chemical oxidation (PCO) and accelerator mass spectrometry (AMS) was pioneered by the Rowe group at Texas A&M University in the 1990s (3). Charcoal pigments are often observed, and are considered more reliable for dating, as charcoal can generally be identified microscopically. Many charcoal images from around the world have been radiocarbon dated with the PCO method developed by the Rowe group (4–7). Applying PCO-AMS to inorganic pigmented paints requires that an organic binder or vehicle was used in the paint, and that the binder/vehicle remains to the present day. Furthermore, because rock painting samples consist primarily of substrate rock, any organic material originating from the substrate must be insignificant. The problem of substrate contamination is the most significant. To address the issue, samples of both paint and unpainted substrate rock have been collected for some studies. We have found that in at least one case in Guatemala, the substrate was heavily contaminated, casting significant doubt on the dates obtained from two other paint samples at that site (8, 9). Paint and substrate samples for that study were not collected systematically. When the Cueva la Conga project was undertaken, we carefully planned collection of matched samples of paint and substrate for a comprehensive analysis to support any attempts at dating.

Materials and Methods Rock Painting Samples In January 2009, two of us (SB and RAA) traveled to Cueva la Conga to collect paint samples in hopes of obtaining radiocarbon dates and identifying the pigments and materials used. Five samples of charcoal and six of inorganic pigmented paints were collected (Table I), all but two of which had unpainted substrates removed as well. Those two samples, from red handprints, were very thick paint that flaked off without significant substrate attached. Figure 1 shows the rock paintings sampled; of particular note is the variety of colors and motifs displayed amongst the paintings. Each sample was removed with a new sterile scalpel blade onto clean aluminum foil that had been baked at 500°C overnight to remove oils; 76 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

they were then wrapped in additional foil and sealed into individual labeled ziptop plastic bags. Samples were documented with photographs before and after sampling.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Microscopy Methods Each sample of paint and substrate was visually inspected at 20x magnification under visible light. Any foreign matter, including fibers and insect parts, was removed prior to further analysis. Two of the samples, 5 and 9, contained charcoal inclusions in the red paints, and were further subjected to scanning electron microscopy. This was carried out by Dr. Glenn Walker of the EMU Biology Department, using an Amray 1820I SEM at 5 kV on gold-coated paint fragments. X-ray Fluorescence Spectroscopy Micro-XRF spectroscopic analyses were carried out in the Conservation Department at the Detroit Institute of Arts. The instrument used was a Bruker AXS ARTAX with a Mo source at 50 keV and 700 µA with no filter. A 0.2 mm collimator was utilized with a helium purge. Acquisition times ranged from 120 to 300 seconds. Thermally Assisted Hydrolysis/Methylation-Gas Chromatography-Mass Spectrometry Gas chromatography-mass spectrometry (GC-MS) has been used to look for lipids—indicative of binder—in rock paintings (10). Pyrolysis-GC-MS has also been used to identify patinas associated with rock paintings and to look for the presence of binders (11–13). We used thermally assisted hydrolysis methylation (THM)-GC-MS, a modified pyrolysis method, to rapidly screen the mineralpigmented paints to determine if binders were present, and to evaluate the efficacy of pretreatment on the charcoal paints. THM-GC-MS is a good method for rapid comparisons of the organic content of small samples like those from Cueva la Conga because the method is fast, requires little sample preparation and is nonselective, yielding results for lipids, amino acids, and carbohydrates at the same time. A Varian 3800 gas chromatograph with a Saturn 2000 mass spectrometer was used for the analyses. The GC was equipped with a Varian Chromatoprobe and 1079 inlet which allowed for temperature programming of the injection port. Submilligram portions of the paint and substrate samples were weighed into Chromatoprobe vials, to which was added two microliters of internal standard-derivatizing agent mixture. This mixture was prepared from 5 µL of a primary internal standard (5 ± 0.01 mg tri-t-butylbenzene in 1 mL of methanol), combined with 1 mL of 10% TMAH in methanol. The vial was placed into the Chromatoprobe and then into the GC inlet. The inlet was cryogenically held at an initial temperature of 40°C for 0.1 minutes, then temperature programmed to 84°C at a rate of 200°C/min and held for 1.00 min at 100% split to evaporate any excess methanol. After this initial 77 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

drying step, the temperature was ramped to 300°C at a rate of 200°C/min; this is the temperature at which THM occurred. Each of the samples was run under splitless conditions. An initial eight minute solvent delay was used at the start of each GC analysis to prevent exposing the MS filament to excess TMAH. Separation was carried out on a Varian VF-5ms column (30 m long, 0.25 mm i.d., 0.25 µm film thickness). Helium (99.999%) was used as the carrier gas, at a pressure of 40 psi. Mass spectra were collected in electron impact mode from m/z 35-650. The ion trap was maintained at 150°C for the duration of the analysis. At the start of each day, a blank consisting of a Chromatoprobe vial with 2 µL of the TMAH containing internal standard was run under the described conditions. Then the samples were run; this allowed us to correct for any contamination in the system. Each sample was run only one time, but the residue was retained for future replicate analyses.

Table I. Description of samples from Cueva La Conga Sample

Location and description

Paint 1

Charcoal from “anthropomorph” figure on Panel 6, (two samples removed)

Substrate 1

Substrate from small pocket in rock to left of anthropomorph

Paint 2

Charcoal from upside-down anthropomorph on Panel 9,

Substrate 2

Substrate from next to Paint 2

Paint 3

Charcoal from fragmentary images in Panel 3, removed from scratches on wall

Substrate 3

Substrate from scratch without charcoal present

Paint 4

Charcoal, possibly from torch smudges, on Panel 7 (not dated)

Substrate 4

Substrate removed from lower vesicle without charcoal

Paint 5

Purple-red paint fragment isolated in vesicle, from red circle on Panel 6

Substrate 5

Substrate from nearby vesicle without paint

Paint 6

Orange paint from complex geometric figure on Panel 4

Substrate 6

Substrate from near orange geometric figure

Paint 7

Red paint from red line spanning Panels 2 and 3

Substrate 7

Substrate from below red line

Paint 8

Yellow paint from handprint underlying red line (#7) on Panels 2 and 3

Substrate 8

Substrate from next to yellow handprint

Paint 9

Red paint flake from handprint in Panel 12

Paint 10

Red paint flake from handprint in Panel 14

78 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Figure 1. Photographs of some of the sampled rock paintings from Cueva la Conga: (A) Sample 1 charcoal anthropomorph; (B) Sample 5 red circle; (C) Sample 2 anthropomorph; (D) Samples 7 and 8, red line and yellow handprint; (E) Sample 9 red handprint.

79 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Plasma-Chemical Oxidation and AMS Radiocarbon Dating The remaining paint and substrate samples were further divided to obtain samples nearly matched in mass prior to wet chemical pretreatment and PCO-AMS. While the standard pretreatment generally consists of an acid wash to remove carbonates, a base wash to remove humic acids, and a final reacidification step to prevent adsorption of atmospheric carbon dioxide (14), we used a modified procedure developed in our laboratory that is less destructive. All samples and substrate samples were weighed into clean microcentrifuge tubes, and combined with 1 mL of phosphate buffer (1 M in phosphate ion, pH (8). Samples were ultrasonicated at 50 ± 5°C for 60 min. Yellow color in the solution was considered indicative of dissolved humic acids; phosphate washes were repeated until the resulting solution was clear following the ultrasonication step. The material resulting from the cleaning step was filtered using glassware and binder-free borosilicate glass filters that had been baked overnight at 500 °C. The material was rinsed with deionized water and dried on the filter. For plasma-chemical oxidation, the filter was placed directly into the plasma-chemical oxidation chamber. The chamber was maintained at a vacuum pressure of ~10-7 Torr. Vacuum integrity checks (VICs) prior to plasma-chemical treatment indicated that no significant leaks were present in the system. We assume, as a worst case scenario, that all pressure increase during the 60-min VIC arises from carbon dioxide; as long as the pressure increase corresponds to less than the contamination background in the accelerator mass spectrometer (typically 0.5-1 µg C), the increase is considered inconsequential. Oxygen gas was of ultra-high purity grade (99.999+ %). The oxygen plasma has been shown to react with organic carbon in the paint samples at a sufficiently low temperature (~150 °C) that the inorganic oxalates and carbonates present are unaffected.

Table II. Pretreatment and plasma conditions Sample

Pretreatment

Oxygen plasma conditions

Yield, µg C as CO2

Paint 1

9x phosphate

50 W, 39 min

260

Paint 2

5x phosphate

50 W, 52 min

90

Paint 3

10x phosphate

50 W, 35 min

275

Paint 5

none

50 W, 41 min

200

Paint 5

2x phosphate

100 W, 60 min

200

Paint 9

none

50 W, 40 min

250

Paint 9

3x phosphate

100 W, 60 min

110

80 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Ideally, 100 µg of carbon as carbon dioxide is preferred to obtain a reliable radiocarbon date. If at least 60 µg C was produced from a paint sample, it was collected by cooling a glass finger on the plasma system with liquid nitrogen. The glass tube was then sealed off and sent to the Center for Accelerator Mass Spectrometry at the Lawrence Livermore National Laboratory for radiocarbon analysis. Pretreatment and plasma conditions for the Cueva la Conga samples are listed in Table II.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Results and Discussion Microscopy The samples that were presumed in the field to be charcoal were inspected at 20x under visible light, and all four of the purported charcoal samples were confirmed to be charcoal. For additional information, samples of the charcoal were selected and sent to Caroline Cartwright, a microscopist and botanical specialist at the British Museum to determine the possible source of the charcoal. Dr. Cartwright is a wood anatomist and has extensive experience with Mesoamerican materials. She has determined that the charcoal pigmented images derive from three botanical sources: Hymenaea courbari, locally known as Jatobá; Pinus species; and Pithecellobium species. The possible cultural implications of the use of these plants are described elsewhere (2). Microscopic examination of the inorganic pigmented samples showed primarily fibers, spider webs, insect parts, etc. Paint sample 7 contained green, algae-like material. Ms. Maria Goodrich examined wet-mounted samples under immersion microscopy at high (400x and 1000x) magnification, and concluded that the material is most likely derived from filamentous cyanobacteria. Cyanobacteria, which probably make up the thick biofilm covering the speleothems at the large entrance to Cueva la Conga, are also common when light levels are very low. Heterotrophic bacteria are ubiquitous in cave environments. This is problematic from the standpoint of radiocarbon dating rock art: significant extraneous organic material in the paint not associated with the painting event will render any date obtained meaningless at best, and misleading in cultural interpretations at worst. Far more promising from a dating standpoint, paint samples 5, 9 and 10 were all observed to contain charcoal inclusions. Pieces of charcoal can be seen protruding from the surrounding mineral pigment, indicating that the charcoal is not superficial, and does not derive from torches or soot. Figure 2 shows an electron micrograph of a charcoal inclusion in paint sample 9. The cellular structure of the wood is clearly visible in the mineral matrix. While there might be cultural reasons to add charcoal to mineral pigment, there are no binding or extending properties that would suggest its use in red or yellow rock paintings. There are further no reports in the literature of charcoal being found in such a manner in rock painting samples. We propose that preparation of the pigments may have led to the charcoal being incorporated into the paint. Changing colors of mineral pigments through roasting is a well-known practice in the art field. An example is the pigment umber, a mixed 81 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

iron-manganese oxide and clay mineral. Raw umber is a light brown color; heating raw umber removes water from the mineral, yielding a warmer, darker brown of burnt umber. This process of calcining iron oxide pigments to change their color has been known since antiquity (15). Because the charcoal inclusions were only observed in the darkest paints from Cueva la Conga, we believe that lighter iron oxide pigments were heated in an open fire to change their color. When the cooled mineral was removed from the hearth and ground to prepare the paint, charcoal from the heating process was incorporated incidentally.

Figure 2. Electron micrograph showing charcoal inclusion in paint from red handprint in Panel 12. Inset shows the black charcoal inclusion in a portion of the paint sample.

X-ray Fluorescence Results The results from the x-ray fluorescence analysis (Table III) support the calcination hypothesis. All of the pigments are primarily iron oxides, with significant contribution from the limestone substrate. Only one sample (5, which was nearly pure paint) did not show significant calcium from the underlying limestone. The titanium and silicon present are likely indicative of sand inclusions. Clays (aluminum phyllosilicates) do not appear to have been added to the paints as an extender, as the aluminum content of the pigments is negligible. 82 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Table III. Elemental compositions of Cueva La Conga paint samples

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Paint sample

Acquisition time, s

Major components

Trace components

5

240

Fe

Ca, Si, Br, Ti (possibly)

6

300

Fe, Ca

K, Ti, Si, Mn

7

300

Fe, Ca

K, Ti, Si, Mn, Zn, Br, As(?)

8

300

Fe, Ca

Ti, Mn, Si, K, Al (?)

9

300

Fe, Ca

Cu, Zn, Mn, Cr, P(?), Br

10

300

Fe, Ca

Ti, Si, Mn

THM-GC-MS Results The primary goal of the THM-GC-MS analysis was to determine whether there were significant differences between the compositions of the paint and substrate samples. Ideally, the organic substances would be found only in the paint samples. Simply put, this was not the case for the samples from Cueva la Conga. The three classes of compounds indicative of binders – proteins, carbohydrates, and fatty acids – are discussed separately. Proteins are cleaved into amino acids and derived into their corresponding Nand O-methyl esters by TMAH. The identity of these compounds was confirmed by running standards of each naturally occurring amino acid under our analysis conditions to build a user database, which was compared to literature spectra for additional confirmation (16). Only four amino acids – alanine, leucine, threonine, and glutamic acid – were observed in any of the samples. When these amino acids were present, they were not isolated in the paint, but present also in the substrate. This strongly indicates that proteinaceous binders were not likely used in Cueva la Conga, or they did not survive to the present day. Carbohydrates are broken into monosaccharides and further converted to methyl ether and ester derivatives during the process. Unfortunately, the resulting compounds are not clearly diagnostic of specific monosaccharides. Instead, multiple compounds that can be characteristic of pentoses and hexoses are formed. Fabbri and Helleur (17) showed that several compounds and their relative ratios can be used to identify monosaccharides. To confirm that these compounds were formed during THM-GC-MS under the conditions we used, we ran seven monosaccharides standards (glucose, fructose, ribose, xylose, galactose, arabinose and mannose) and found that each monosaccharide does indeed yield several peaks, making the identification of any single sugar complex. A few compounds were selected as markers for sugars, including 2, 4-dimethoxybutanoic acid methyl ester and the permethylated saccharinic acids. The paint and substrate pairs for samples 6 and 7 were of the same composition; for sample pairs 5 and 8, two of the carbohydrate markers were found in the substrates while only one was observed in the paints. This indicates that carbohydrate-based binders either were not used or did not survive at Cueva la Conga. 83 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

Figure 3. Graphs showing the comparative fatty acid composition of (A) Sample 5 and (B) Sample 7.

The compositions of the paint samples were dominated by saturated and unsaturated fatty acids ranging from C8:0 to C18:0, as their methyl esters. The most common fatty acids identified were saturated C9 and C17, even-carbon chains from C10 through C18, and monounsaturated C18. Nonanedioic acid, present as the dimethyl ester, was identified in both the paints and substrates of paint samples 7 and 8. Decanedioic acid, also as the dimethyl ester, was detected in both the 84 In Collaborative Endeavors in the Chemical Analysis of Art and Cultural Heritage Materials; Lang, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2012.

Downloaded by STANFORD UNIV GREEN LIBR on July 17, 2012 | http://pubs.acs.org Publication Date (Web): July 10, 2012 | doi: 10.1021/bk-2012-1103.ch004

paint and substrate of paint sample 7. These compounds are formed from the decomposition of C18 and C20 fatty acids, respectively. Fatty acids were also observed in blank runs, so a semi-quantitative analysis of each run was carried out using the mass of the internal standard. The mass of each compound was then corrected for the mass of sample used, and the blanks were subtracted from each sample. The corrected amount of each fatty acid derivative was plotted on a bar graph for comparison of the paint and substrate compositions. Figure 3a shows the fatty acid bar graph for sample 5; the graphs for samples 6 and 8 are generally identical. With a single analysis, no error bars can be drawn, but the general pattern of similarity between the fatty acid composition of the paint and substrate is readily observed. In the case of the paint and sample pair 7, a different trend is observed (Figure 3b). This may indicate that something like animal fat, now heavily degraded, might have been used as a binding medium by the artists of Cueva la Conga in the case of paint sample 7. In order to test this hypothesis, we compared the fatty acid composition of that sample to that of an animal fat prepared as a paint for another study in our laboratory several years ago. We can only reliably say what the samples are not. It is impossible to test every possibile binder, so positive identifications are not something that can be made definitively. Differences are more significant than similarities. For example, many more dicarboxylic acids were observed in the animal fat in large quantities, yet were not observed in any significant amount in paint 7.

Table IV. Ratios of fatty acids for distinguishing food residues (18) compared to animal fat standard and Cueva la Conga paint 7

Ratio

Degraded fat from terrestrial mammals

Animal fat standard

Paint 7

Degraded fat from fish

(C15:0+C17:0)/C18:0